Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Multi-Agent Systems for Resource Allocation and Scheduling in a Smart Grid

Multi-Agent Systems for Resource Allocation and Scheduling in a Smart Grid With the increasing integration of Distributed Energy Resources (DER) in the power grid, a decentralized approach becomes essential for scheduling and allocation of resources in a smart grid. Economic Dispatch (ED) and Unit Commitment (UC) are the two major resource allocation problems that play critical role in the safe and stable operation of a grid system. The uncertainty associated with renewable energy sources have made the resource allocation problems even more challenging for grid operators. The future grid will have a higher generation mix of renewable energy sources and a large load of Electrical vehicles, with the possibility of bi-directional power flow. This complex smart grid system necessitates the development of a decentralized approach to resource allocation problem, which allows inter-node communication and decision making. Multi-agent systems (MAS) is a promising platform to decentralize the traditional centralized resource allocation aspects of smart grid. This paper presents a comprehensive literature review on the application of MAS to Economic Dispatch (ED) and Unit Commitment (UC) in smart grids. . . . . Keywords Multi-agent systems Smart grid Resource allocation and scheduling Economic dispatch Unit commitment Introduction The presence of increased penetration of Renewable Energy Sources (RES) in the power system creates many The smart grid framework incorporates distributed generation, technological challenges for the power companies owing to advanced communication network, smart meters and sensors the need for improved system control to maintain the power to make the grid more reliable, flexible, adaptive and efficient. quality to consumers [3]. The mix of conventional and RES This new power system paradigm necessitates the need to must work in tandem to maintain the power generation at the revisit some of the traditional power system operations to meet required level. The process of committing the generators and the challenges of next-generation transmission and distribu- allocating the required generation level has become a chal- tion systems [1]. Along with the integration of renewable en- lenge to meet the increased demand, while using the genera- ergy sources, the deregulation in the energy market has creat- tion from uncertain RES. These factors make the centralized ed competition for power generation companies. Generation control of a smart grid system complex and less efficient to companies have an obligation to meet the customer energy process the diversity of data and controls [4]. The concept of demands even during peak hours and system outages. There Multi-Agent Systems (MAS) is put forward to solve this prob- is a need to properly allocate the generation sources to maxi- lem by using automated agent technology. The MAS converts mize the profit considering renewable generation and custom- a centralized control system into a distributed control model at er demand [2]. a component level. MAS is a collection of agents working together with each other to achieve an overall objective [5]. An agent can be defined as a computer system with the ability to take critical * Arun Sukumaran Nair arun.sukumarannair@und.edu decisions based on the scenario to improve its objective [6, 7]. These software agents in a smart grid environment sense, 1 communicate, collaborate and act with each other. The agents Department of Electrical Engineering, University of North Dakota, can act autonomously or semi-autonomously, with local or Grand Forks, ND, USA 15 Page 2 of 15 Technol Econ Smart Grids Sustain Energy (2018) 3:15 global information [8]. MAS technology is finding wide range The centralized MAS architecture for micro grid control is of applications in the power system domain such as optimal discussed in the literature [34, 35]. The framework of distrib- power flow [9, 10], power system restoration [11–17], elec- uted and three level hierarchical MAS for a smart grid is ex- tricity market operation [18–21], power system control plained inP. Luetal. [36] and K. E. Nygard et al. [37] [22–26] and protection [27–29]. The focus of this paper will respectively. be restricted to the application of MAS in the fundamental Most of the MAS based optimization problems rely on the resource allocation aspects of the power grid namely, technique of consensus algorithm to reach the solution. The Economic Dispatch (ED) and Unit Commitment (UC). main idea behind the consensus problem is to make a set of ED is one of the most important challenges in the pow- agents agree up on a certain value (usually a global function) er systems and it deals with the allocation of power gen- by using local information exchange among agents (local in- eration among committed generators in order to meet the teraction). This concept is utilized in different fields such as demand while lowering the generation cost [30]. economics (Agreement problem), communication (Gossip al- Consumer demand for clean energy and government reg- gorithms [38]), statistical mechanics [39] (Synchronization) ulation has motivated the integration of Distributed and robotics (Flocking [40]). It offers several advantages over Energy Resources (DERs) like solar photovoltaic, wind traditional centralized methods such as distributed computa- power, and fuel cells into the modern power grid. This tion, computational efficiency, independent of graph topology makes ED a highly complex optimization problem which and robustnesstofailure [41, 42]. In a consensus algorithm needs to consider the various factors like generator capac- model, each node in the system is considered as a dynamic ity, ramp-rate, failure rate, emission, load profile and gen- agent with a value or state associated with it. The value of the eration from DER. Unit Commitment (UC) in a smart grid agent represents the decision variable with which it can reach system is a highly complex optimization problem that consensus with other agents in the system. Researchers have schedules the startup and shutdown of generators to meet explored different census methods for a microgrid system the demand while satisfying system constraints [31, 32]. such as in the work by G. Hug et al. [43] where a combination The committed generators are modeled in the ED for gen- of consensus and innovation method was utilized. A novel erator scheduling. The smart grid systems which have framework to model a full automation of a distributed smart significant DER and the increased interest from con- grid system is presented in the work by K. E. Nygard et al. sumers to install RES have necessitated the need for a [37]. The model is based on the concept of an Intelligent decentralized approach to commit and schedule genera- Autonomous Distributed Power System (IDAPS), a microgrid tors. The increased uncertainty from RES has made ED with sufficient resources and intelligence to function autono- and UC more complicated due to the intermittent nature mously within a global grid. A three-layer hierarchical system of these power sources. model with agents in higher level supervising the agents in For understanding the application of MAS in resolving lower levels is proposed. The model accomplishes modularity, resource allocation and scheduling problems of smart grid, scalability, and a balance between global and local decisions the paper is organized in five sections. Section II describes of agents. Distributed MAS based control offers several ad- the architecture of a MAS in a smart grid system, section III vantages such as autonomy, fault-tolerance low latency, effi- and IV present comprehensive review on the application of ciency and much more. It is a way of physically breaking MAS in ED and UC respectively, followed by conclusion in complex control problems into smaller control problems, section V. and then solving them closer to the control operation itself. Development of Smart Grids will involve dealing with a bigamountofdatacollectedinadistributedmanner.Thisdata is communicated among equipment and devices to support MAS for Smart Grid decision-making process. Certainly, handling the amounts of data to be acquired and processed in such distributed systems An agent represents a computer system situated in an environ- to extract useful information bring its own challenges. ment where it is capable of making decisions to achieve its Computational intelligence techniques are used to extract design objectives. Moreover, an agent can be autonomous, knowledge and overcome some of the challenges. In “intelli- social, reactive and proactive. Multi Agent Systems (MAS) gent” systems, data is pre-processed, processed, and then are composed of agents interacting in a highly dynamic envi- information is extracted for decision-making. Given the dis- ronment. These intelligent agents are being developed to have tributed nature of Smart Grids (SG), new advancements in the functionalities on par with the human experts to act appro- distributed intelligence techniques spawned the MAS tech- priately in the various scenarios that take place in a smart grid. nique development. The following are some advantages and The summary of various MAS architectures used for control reasons to explore distributed MAS architectures over cen- tralized architecture: of microgrid is summarized in Table 1 [33]. Technol Econ Smart Grids Sustain Energy (2018) 3:15 Page 3 of 15 15 Table 1 Different MAS architectures for microgrid control [33] MAS architecture Type of agent Role Features Centralized Cognitive Agent Higher level of intelligence/ + collects information communication capabilities at asinglepoint Reactive Agent Fast Response + capable of making global decision + flexibility and openness in the operation of smart grid − suffers from computational burden in case of large number of agents − single point of failure affects the entire system Two-level hierarchical High level agent Infrastructure management, + distinct levels of decision making low level scheduling − failure of higher level agents results Low level agent Accept schedule from High in critical conditions of the lower level agents level agent, asset management Three-level hierarchical High level agent Critical decisions, data and + good scalability through delineation policy management of roles to agents Middle level agent Fault location, switching of grid − failure of higher level agents results in connected/islanded mode critical conditions of the lower level agents Low level agent Sensor management Distributed Local Agent Local information discovery/ + robust system with agents being capable communication of reorganizing and coping up with the loss of other agents Advantages of Distributed MAS over centralized Response (DR) with the ability to dynamically optimize control: grid operations , resources and consumer participation. To do so, it is important to understand demand participation 1. The SG components are often distributed and the of consumers. As number of consumers are growing, it is energy management system is tightly associated with essential to do the demand response analytics in a distrib- the communications between stakeholders and enti- uted fashion. ties (agents) to exchange information, so MAS is an appropriate platform to develop distributed manage- Due to the inherent advantages of distributed MAS archi- ment functions. tecture, it is well suited to resolve the complex ED problem of 2. SG is a holistic system and the failure of some part of it smart grid. A general framework for the implementation of (e.g., the breakdown of a transmission line or cut down of MAS in smart grid (SG) system is shown in Fig. 1 [44]. SG a substation, transformer) should not affect the whole ac- system is an integration of the physical grid with the commu- tivities and operations, and hence fault tolerance can be nication layer where the agents act as an interface. The com- easily attained in distributed architecture over centralized munication layer is a strongly connected network with varying architecture. and configurable topologies. Each agent can be categorized 3. SG should demonstrate the plug-and-play concept for in- into three units; namely Device Unit (DU), Decision Making tegrating energy storage, loads, and sources at the build- Unit (DMU) and Communication Unit (CU). DUs can be ing level with the external utility grid. Plug and play considered as physical power system buses with components adaptability is widely proven by MAS. The nature of such as Synchronous Generators (SGR), Renewable MAS enables it to scale up by adding other agents or by Generators (RG), flexible load and rigid load. DMUs perform dispersing them in new environment with new resources the local computing for the agents and CUs are the communi- and capacities. Hence, Distributed MAS building mod- cation nodes, which transmit and receive information [45]. ules are highly scalable, and modular. The internal structure of an agent model is shown in Fig. 2. 4. As SG will be composed of an aggregate of Micro grid, An agent model consists of three units; Communication Unit and hence the control can be delegated to micro grids. (CU), Decision Making Unit (DMU) and Device Unit (DU). With futuristic smart grids being a simple collection of CUs are generally signal receivers/transmitters used to ex- residential microgrids, each microgrid can exhibit distrib- change information with neighbors. The calculator, sensors uted control. and controllers are part of DMU, which are responsible for 5. One of the goals of the SG is to develop grid moderniza- the local computing in an agent. DMU is the brain of an agent tion technologies, tools, and techniques for Demand- node and capable of generating control instructions for the DU 15 Page 4 of 15 Technol Econ Smart Grids Sustain Energy (2018) 3:15 Fig. 1 Framework for distributed multi-agent system for smart grid Decentralized Communication Network [44] CU CU CU DMU DMU Agents DMU SGR RG Load Physical Grid as well as responsible for communicating the information to generators in the system to meet the demand. ED needs to the CU. DUs represent the traditional buses in a network conform to several other constraints for a safe and secure which consists of different elements such as synchronous gen- operation of the grid. The integration of uncertain renewable erators, Renewable Generators (RG), battery storage systems energy sources to the grid has made ED and power quality (BESSs), flexible and rigid loads. DU performs the control analysis more important and also more complicated [52, 53]. suggestions from DMU and also sends the feedback to the Distributed algorithms are becoming popular for intelligent DMU. decision-making and control and these algorithms appear to There are number of simulation and open source tools be promising in the context of smart grid. These algorithms available for modeling MAS platforms [46]. The most com- are robust, immune to topological variations and can support mon ones are ZEUS, AgentBuilder, JADE, and MADKit. the “plug-and-play” feature of the future grid. However, it is Features of these MAS modeling tools are listed in Table 2. more challenging to include the operational constraints in such a distributed formulation. Many researchers have proposed a consensus-based approach for ED without losses and lower Economic Dispatch and upper power boundaries. A consensus algorithm is widely used in solving the ED problem in smart grid. It is a method Economic Dispatch (ED) is one of the fundamental problems used to achieve agreement on a single data value among dis- in the power system domain. It is basically an optimization tributed systems. This algorithm is designed to achieve reli- problem with the objective of reducing cost while maintaining ability in a network involving multiple unreliable nodes. [36, the generation-load balance. ED schedules the committed 54–56]. Fig. 2 Internal structure of an Agent i agent [44] in out Information Communication Unit (CU) Exchanging clock, signal receiver & transmitter External Internal Information Information Local Decision Making Unit (DMU) Computing monitor, sensor, calculator & controller Sampling Control signals suggestions Device Unit (DU) Power synchronous generators/RG/BESS/load Flow Technol Econ Smart Grids Sustain Energy (2018) 3:15 Page 5 of 15 15 Table 2 MAS modeling tools MAS tools Description ZEUS ZEUS [47] is a multi-agent platform developed by the research program of British Telecom intelligent system research laboratory. ZEUS allows the design of multi-agent distributed systems. This platform, developed in Java, automatically generates Java code from the agents specified graphically. Agent Builder Developed by Reticular Systems Inc., AgentBuilder [47], it is based on BDI (Believe - Desire - Intention) models Agent [48] and AGENT-O language [49]. It is remarkable for the quality of its software and a good academic model. AgentBuilder is a commercial design software for “intelligent” agents i.e., cognitive and collaborative agents. AgentBuilder consists of two main components: the toolkit and runtime system. JADE JADE [50] is a multi-agent (multi-host) platform developed by Bellifemine. F., Poggy. A., Rimassa. G. and P. Turci by Telecom Italia Lab “Tilab formerly CSELT” in 1999. This platform aims to simplify the construction of interoperable MAS, achieve compliant applications with the standard FIP A97 (Foundation for Intelligent Physical Agents) to facilitate the communication of JADE agents with non-JADE agents, and optimize the performance of a distributed system agent. JADE includes all accredited component that manages the platform: Agent Communication Channel (ACC), Agent Management System (AMS), and Director Facilitator (DF). MADKit MADKit [51] is a platform for MAS developed by Olivier Gutknecht and Jacques Ferber in Laboratory of Computer Science and Robotics and Microelectronics of Montpellier. MADKit was motivated by the need for a more flexible platform possible, and able to adapt to different agent models and application areas. MADKit is a modular multi-agent platform and scalable, written in the Java language. It allows the creation of MAS based on the relational model Aalaadin or AGR (Agent / Group / Role): agents are located in groups and play roles. MADKit takes advantage of object-oriented programming: MADKit features are contained in the MADKit kernel. In a conventional centralized method (e.g. Lagrange mul- to acquire system power mismatch. This new approach will tiplier method), at the optimal point, all the generators will eliminate the need for a single leader node to do all the calcu- have the same incremental cost. An appropriate consensus lation. An average consensus will run at the lower level of the algorithm can guarantee a similar result by having all the two-level method and ICC will be employed at the second consensus variables to converge to a common value asymp- level to process the mismatch information. This is an im- totically. Based on this concept, Z. Zhang and M. Chow [57] proved version from the method proposed in [57]. It is more introduced an Incremental Consensus Algorithm (ICC) to de- distributed and does not require a fixed communication centralize the ED problem by choosing incremental cost (IC) network. as a consensus variable. The model consists of a local control- A decentralized approach to ED in a microgrid with ler (generation unit) which will update its consensus variable Distributed Generators (DG) was explored by N. Cai et al. depending up on the neighbor’s values. The proposed ap- [7] using a MAS. Here, each DG was assumed to have an proach requires a leader node, which will decide whether to agent which could receive local information and communicate increase or decrease the IC based on the demand constraints. only with its nearest neighbors. Agents compete with others to The authors tested the approach on a 3 unit and 5 unit system obtain a local solution, thereby obtaining a global optimum. to test the validity and convergence of the proposed approach. The authors used the concept of consensus among agents to They showed a successful implementation of consensus algo- obtain the optimum solution for the ED. The authors validated rithm in ED but lacked a fully distributed model since it need- the approach on five and fifty agent systems but did not com- ed a leader node to control the agents. A more detailed de- pare the results with a centralized approach. scription about ICC is provided in [58]. A consensus control based approach to solve the ED prob- The previous paper utilized ICC algorithm to implement lem in a smart grid was developed by S. Yang et al. [60]. The ED in a distributed fashion but relied on a leader-follower approach solves the ED in a distributed fashion with the gen- consensus algorithm. A leader node needs to be selected, erators acting collectively to receive the mismatch between which will gather the local power mismatch from the follower demand and power generation information, which is the feed- nodes to calculate the total power mismatch. The follower back for the agents. The total mismatch is generated in a col- nodes need to report their power mismatch to their leader. Z. lective fashion from the estimate of local mismatch by the Zhang et al. [59] introduced a two-level consensus approach agents, which removes the need for a leader agent to collect 15 Page 6 of 15 Technol Econ Smart Grids Sustain Energy (2018) 3:15 the global information. The incremental cost of the generators F. Guo et al. [65] explored the potential of a distributed ED is chosen as the consensus variable and incremental cost cri- model for a smart grid system with random wind power. The terion was used to obtain the optimal dispatch. The method proposed model works on the projected gradient and Finite- was found to have the same precision as the Lambda-Iteration time Average Consensus Algorithm (FACA) and supports the approach, a centralized method, with less communication plug-and-play feature of new generation smart grids. The ran- overhead. dom wind power generation is modeled using the determinis- A distributed ED model considering line loss was devel- tic method with overestimation and underestimation cost var- oped by G. Benetti et al. [61]. The nodes in the model run two iables. The agents can choose arbitrary initial values and are consensus methods in parallel; one to find the Lagrangian not required to share gradient or incremental cost information variable and a second one to find the power mismatch. The with the neighbors. The authors validated the effectiveness first method is a first-order consensus algorithm which uses a andperformance of theproposedmethodonIEEEtest proportional controller to bring the power mismatch to zero systems. and to satisfy the generation-demand equality constraint. The A consensus-based distributed ED taking into account gen- second consensus method uses the work allocation concept to erator dynamics was studied by J. Cao et al. in [66]. The find the power mismatch. The authors assert that the proposed authors used comprehensive generator constraints to improve method can satisfy generation constraint and can handle line the consensus algorithm and analyzed the effect of different loss in the system. The comparison between the distributed communication topologies on the speed of the consensus al- approach and the centralized approach to verify its conver- gorithm. The model relies on local power mismatch data from gence speed and accuracy was not attempted by the authors. the agents rather than a leader node to collect global informa- A. Cherukuri et al. [62] explored the concept of distributed tion. The authors asserted the superiority of the proposed consensus-based approach to model an ED which can handle method by comparing with Lambda iteration and PSO changing load conditions and can remain stable under inter- methods. The generator dynamics was found having a signif- mittent power sources. The proposed model employs two dy- icant effect on the speed of the consensus method while the namical systems namely, dynamic average consensus and effect of communication topology was not significant. Laplacian non-smooth gradient. The mismatch between gen- A distributed consensus-based approach to solve ED in a eration and load is estimated in a distributed fashion by the microgrid was developed by Z. Yang et al. [67]. They used a consensus method and the Laplacian non-smooth gradient dy- novel concept of virtual incremental cost as the consensus namically allocates the generation. The approach can reach variable which does not require the nodes to share power optimum solution from any initial power allocation and does output or generator parameters. The algorithm has the advan- not require a feasible allocation as the initial value. The au- tage of not depending on the local power mismatch to reach the optimum and maintaining the supply-demand balance thors verified the effectiveness of the method to handle dy- namic loads and intermittent power sources. even during transients. They reported reduction in communi- K. Luo et al. [63] developed a MAS based distributed ED cational burden between nodes and improved reliability of the model for an electrical grid system with RES, which can be algorithm. deployed for real-time applications. The proposed approach is Y. Li et al. [68] developed a distributed ED model for a a two-step process, with the first step calculating the initial combined heat and power system. The MAS based framework generation values using adjacency average allocation algo- utilized two consensus protocols, one optimizes the electrical rithm and the second stage performs the ED in a distributed incremental cost function while the other gets a common value manner using local replicator dynamics. The first stage han- for the heat incremental cost. The heat and power coupling in dles the equality constraints in the model while the second the objective function and constraints are managed by these stage conforms to the inequality constraints. They validated two consensus variables. It works in a completely distributed the effectiveness of the proposed method but did not compare fashion without the need for a leader agent with the global the performance of the method with similar approaches. information. They report the effectiveness of the proposed A distributed ED model for an islanded microgrid system ED model by comparing to a centralized approach using was developed by P.P. Vergara et al. in [64]. The model con- Lagrangian relaxation method. sidered both active and reactive power in the optimization Z. Yang et al. [69] proposed a distributed consensus-based model. The primal-dual constrained optimization method model for the ED in a smart grid system which maintains the was used to solve the problem in a distributed fashion, in supply-demand balance even during the transient process. The which two consensus methods are executed in parallel to ob- method has the advantage of not relying on the supply-demand tain the dual values or incremental costs. The authors validat- mismatch and hence can be used online. The proposed method ed the performance of the proposed method by comparing to a does not require a leader node with the complete information of classical Lambda method and also the capability of the meth- power demand in the grid system. It uses the maximum incre- od for fault tolerance. mental cost of the neighboring generators and developed a Technol Econ Smart Grids Sustain Energy (2018) 3:15 Page 7 of 15 15 method to increase or decrease the incremental cost of a satu- A dynamic agent-based approach to model a decentralized rated generator to maintain the supply-demand balance during ED was developed by V. Loia et al. [76]. ED was solved using iterations. H. Xing et al. [70] utilized an average consensus self-organizing dynamic agents equipped with distributed based bisection approach to perform distributed ED. The meth- consensus method. C. Zhao et al. [77] explored the effect of od has the advantage of not relying on prior information or a cyber-attacks on a consensus-based ED model. The authors leader node to perform the optimization. tested the performance of the algorithm for false data injection G. Binettietal. [71] developed a distributed model to into the broadcast information, offline and online ED models, solve non-convex ED problems. The non-convexity comes and bounded and unbounded generation cases. from the valve-point effect, prohibited operating zones, The increased amount of communication between nodes in multiple fuel option and transmission losses but makes the a smart grid system can lead to communication bottlenecks modelmorerealisticfor real-timeoperations. Theproposed which can cause convergence issues in consensus-based ED model is fully decentralized and does not require a leader models.C.Lietal. [78] developed an event triggered node with the global information. The method has the added consensus-based ED model to reduce the communication advantage of being deterministic while heuristic methods do overhead in a smart grid system. The authors reformulated not guarantee the uniqueness of the solution from a single the ED model using θ-logarithmic barrier to conduct the in- run. A combination of auction mechanism and market- formation exchange in a distributed fashion. The reformulated based MAS was used to design the distributed ED and the ED model is solved in a two-stage process; the initial values authors tested the validity of the method on standard test for the agents are generated using connected dominating set systems. G. Binetti et al. [72] also proposed a distributed based distribution algorithm as the first stage and in the next ED model which considered transmission losses in the sys- stage a consensus-based optimization is applied to the system. tem using a combination of two consensus algorithms run- The authors stated that asynchronous communication-based ning in parallel. The model utilized a first order consensus event triggered ED model can significantly reduce the com- protocol to calculate the local power mismatch to satisfy the munication exchange in a smart grid system, but the event demand constraint and a second consensus algorithm to cal- triggered mechanisms can have a negative impact on the con- culate the system power mismatch. vergence rate. A fast gradient-based method is used to accel- A transition of the MAS based distributed ED from labo- erate the convergence rate in the optimization model. ratory set up to industrial model is studied by G. Zhabelova Most of the papers discussed above assume a perfect com- et al. [73]. An incremental cost consensus approach model munication between agents without any information loss, but based on the industrial standard IEC 61499 is used to solve in a realistic smart grid environment can have packet loss and ED in a smart grid environment. IEC 61499 is a promising communication failures. Y. Zhang et al. [79] proposed a dis- tributed ED model which remains robust under information industrial standard used as an architecture for the development of distributed systems in control and automation. The agent- loss among agents. A combination of two algorithms running based system modeled after the IEC 61499 standard will be in parallel, Robust distributed system Incremental Cost suited for industry application and can be executable on the Estimation (RICE) algorithm, was introduced by authors to target platforms. The authors tested the proposed model on a handle the issue. The model contains a Gossip algorithm to 5-node system with industrial controllers. find the power mismatch estimation and consensus algorithm A combination of MAS and Particle Swarm Optimization for the incremental cost estimation. They report that their (PSO) called MAPSO (Multi-Agent Particle Swarm method outperforms the consensus method to packet loss Optimization) was proposed by C. Wu et al. [74] and was and delivered good results even with a 5% information loss applied to the ED problem. The proposed method overcomes in the network. Another study on distributed ED under com- the shortcomings of PSO, the fast convergence to the local munication uncertainties is by G. Wen et al. [80]. The pro- optima, and achieves high convergence speed and precision. posed approach utilized a robust consensus model to counter The agents are modeled to have the ability of self-learning to the communication uncertainties. A study of consensus-based improve the problem-solving ability. The authors verified the ED model under dynamic communication network is evaluat- effectiveness of the proposed method on IEEE test buses and ed by M. Hamdi et al. in [81]. T. Yang et al. [82]explored a the method was found to be faster than evolutionary algo- distributed ED model for a system with potentially time- rithms. A hybrid of MAS with PSO, deterministic search varying topologies and network delays. The authors proposed and bee decision-making process called HMAPSO (Hybrid a gradient push-sum based method to handle the network Multi-Agent based Particle Swarm Optimization) was pro- challenges. posed by R. Kumar et al. [75]. The HMAPSO method was The application of distributed ED using consensus the- applied to an ED model with valve-point effect and was ob- oryinamicrogridisbyproposed R. Wang et al.[83]. The served to be more robust and accurate than other PSO proposed method is a fully distributed approach without a methods. leader or a virtual control node. The incremental cost of 15 Page 8 of 15 Technol Econ Smart Grids Sustain Energy (2018) 3:15 each bus in the system is taken as the consensus variable. communication in the system. The decrease mobile agent The authors validated the performance and convergence (DA) are intended to reduce the generated output and of the distributed ED in a microgrid model. A similar increase mobile agents (UA) initiates an increase in gen- approach for distributed ED in an islanded microgrid is erated output. In the proposed approach, mobile agents proposed by Z. Tang et al. [84]byusing IC as thecon- travel throughout the system and negotiates with the gen- sensus variable. A distributed power dispatch model for a erator agents, depending up on the operating conditions. multi-microgrid scenario is reported by X. He et al. [85] The performance comparison of the method with dynamic utilizing a primal-dual consensus algorithm. They evalu- programming yielded similar results but it is not a fully ated the performance of the proposed algorithm on an decentralized method since it has a leader and mobile IEEE 30, 57 and 300 bus systems. A study on distributed nodes as an interface between generator agents. Figure 4 control architecture for a hybrid AC/DC microgrid is per- shows the architecture of the proposed method. formed by P. Lin et al. in [86]. In most of the above cited An improved version of the method proposed in [31] papers for ED, an evaluation of generation cost is not was presented by J. Yu et al. in [88]. The proposed MAS performed for the different approaches and moreover the agents are more intelligent and have the capability to problem of ED is not solved in a more realistic manner solve complex optimization problems. The profit maximi- with non-negligible losses. Limited work has been done zation objective is obtained using three types of agents, on the implementation cost of these approaches in a smart namely central agent, mobile agent and generator agent. grid system. A summary of the above cited papers and The central operator in the system is the central agent their features are listed in Table 3. which commands the mobile agents to achieve the objec- tive function. The mobile agents travel to each generator to negotiate and reach a satisfactory result. The model is Unit Commitment not fully decentralized as there is an agent acting as a central controller. The authors validated the proposed Unit Commitment is the process of determining the schedule model against a hybrid Lagrange Relaxation - Evolutio of generating units within a power system. The optimized nary Programming (LR - EP) method. schedule is generated subject to device and operating con- A MAS based UC model for a smart grid system consid- straints of the system with the objective of minimizing the cost ering RES uncertainty was developed by X. Zhang et al. [89]. for utilities [92]. The ED optimization is usually performed on The model considered the uncertainty associated with wind, the committed generators from the UC step. Various ap- solar and load along with the charging and discharging of proaches were used to find the optimal schedule from the PHEVs (plug-in hybrid electric vehicles). The hierarchical system consisted of management agents, work agents, coop- UC problem ranging from highly complex and theoretical methods to simple rule of thumb methods [93–97]. The scope erative co-evolution agent and object cooperative agent. The of the UC problem depends on the generation mix, operating work agents used adaptive GA to solve the optimization. and security constraints set by the utility. The focus of this A MAS based approach to solve the profit based UC review is on the decentralized approaches which utilized problem was explored by J. Yu et al. [99]. Rule-based, MAS to solve UC problem. and dynamic programming methods were used to solve Authors in [87] developed a centralized approach to solve the profit based UC. D. Sharma et al. [91] introduced an UC in a smart grid system using a MAS based architecture. improved version for the profit based UC. The ISO agents The agents communicate information to neighboring agents, in the proposed method used a rule-based intelligence to but the UC happens in a centralized controller. Figure 3 shows work in conjunction with generator agents to maximize the different agents in a centralized UC. The proposed method the objective function. The functionality of generator helps to reduce the communication overhead but has the de- agents is limited to maximize their profit for a given de- merits of a centralized controller such as a single point of mand and reserve using real – parametric Genetic failure, increased computation time with complexity, unavail- Algorithm and to share the information with the ISO ability of plug-and-play functionality etc. agents. The maximum profit generating agents are com- A distributed UC model based on MAS was developed mitted to the system by ISO agents while satisfying the by T. Nagata et al. [31, 98]. The proposed model consists up/down time constraints. The authors reported the per- of three types of agents namely Generator Agents (GA), formance of the proposed method with several hybrid mobile agents (DA, UA) and Facilitator Agent (FA). The methods. FA contains the objective function. The system level con- T. Logenthiran et al. [2] utilized MAS concept to de- straints are satisfied by the interaction between GA and velop a resource scheduling model for an islanded power mobile agents and the GA handles the local constraints. grid with integrated microgrids and DER. The proposed methodology has three stages; microgrid scheduling to The two mobile agents are provided to improve the Technol Econ Smart Grids Sustain Energy (2018) 3:15 Page 9 of 15 15 Table 3 High-level systematic summary of MAS application in ED and UC Problem/ Algorithm Platform Constraints Implementation Author, Year Remarks Architecture ED/Distributed Incremental Cost Consensus (ICC) – Demand Constraint Simulation on 3 unit and 5 unit Z. Zhang et al. + Successful implementation of system [57] consensus algorithm in ED 2011 - Leader node need to be selected Average consensus + ICC – Demand, Generator constraints Simulation on a 5 unit system Z. Zhang et al. + Not a leader-follower structure [59] + Computation more distributed Consensus based – Demand, Generator constraints 5 generators in a prototype N. Cai et al. [7] + utilized Local information among microgrid 2012 agents + proposed an improved communication algorithm for the agents Consensus based MATLAB Demand, Generator constraints Simulation on IEEE 14 bus S. Yang et al. + utilized mismatch between demand and system [60] power generation 2013 + same precision as the centralized method (Lambda-Iteration approach) Incremental Cost Consensus (ICC) MATLAB Demand Constraint 5 Node system (IEC 61499 G. Zhabelova + Developed an industrial model for the Architecture) et al. [73] MAS based ED 2013 + used industrial standard IEC 61499 Consensus based-parallel MATLAB Demand, generation, Transmission loss Simulation on IEEE 6 and 300 G. Benetti et al. +usedtwo consensus algorithms in constraints bus system [61] parallel 2014 + communication network has the same topology of the power system Dynamic average Consensus – Power and Generator constraints 15- Bus System A. Cherukuri + improved distributed coordination et al. [62] algorithm 2014 + can handle intermittent energy sources, dynamic load conditions Consensus based MATLAB/Simulink Generator, ramp-rate limit, line-flow Simulation of a five-generator J. Cao et al. [66] + considered generator dynamics limit constraints smart grid model 2014 + compared different communication topologies Distributed auction-based MATLAB Valve point effect, multiple fuel options Simulation on 10, 15 and 40 G. Binetti et al. + ability to solve non-convex ED algorithm and prohibited operating zones generators [71] problems 2014 +Utilizes a leaderless consensus protocol +More robust & Fault tolerant Distributed average consensus DICE /MATLAB Demand, Simulation on IEEE 118 and V. Loia et al. + self-organizing dynamic agents Generator limits 300 bus system [76] 2014 + A prototype version of the self-organizing architecture is developed Distributed replicator dynamics JADE Demand, Simulation on system with 3 K. Luo et al. + can be deployed for real-time Generator constraints DGs and 3 loads [63] applications 2015 + Fully distributed computation + utilized game-theory concepts Multi-Agent PSO MATLAB Demand, Simulation on IEEE 3, 13 and C. Wu et al. [74] + used a combination of MAS and Generator constraints, Valve-point effect 40 units 2015 Particle Swarm Optimization + faster than evolutionary algorithms Projected Gradient and Finite time – Demand, Generator constraints Simulation on 6 bus and IEEE F. Guo et al. + considered random wind power average consensus 30 bus system [65] generation 2016 + Incremental cost of generators not required 15 Page 10 of 15 Technol Econ Smart Grids Sustain Energy (2018) 3:15 Table 3 (continued) Problem/ Algorithm Platform Constraints Implementation Author, Year Remarks Architecture + Plug-and-Play property Consensus based – Demand, Generator constraints Simulation on a Five-unit Z. Yang et al. + virtual incremental cost as the system [67] consensus variable 2016 + maintains the supply-demand balance during transients. Consensus based- two protocols – Electrical and heat power balance Simulation on 16 Bus test Y. Li et al. [68] + modeled ED for a combined heat and constraints, capacity limits system 2016 power system + No leader or central controller Consensus based – Demand, Simulation on IEEE 57 Bus C. Li et al. [78] + event triggered consensus-based ED Generator constraints system 2016 model + reduces the communication burden in the network Robust distributed system – Demand, Simulation on IEEE 9 Bus Y. Zhang et al. + combination of two algorithms running incremental cost estimation Generator constraints system [79] in parallel (RICE) 2016 + can remain robust under information loss Consensus based- two in parallel – active and reactive power balance Simulation on IEEE 30 Bus P.P. Vergara + considered both active and reactive constraints system et al. [64] power in the model 2017 + based on primal-dual constrained decomposition Consensus based – Demand, Simulation on IEEE 14, 118 Z. Yang et al. + Does not rely on the supply-demand Generator constraints Bus systems [69] mismatch 2017 + Can be used in real-time applications Consensus based – Demand, Generator constraints Simulation on 10 bus and IEEE R. Wang et al. + Implemented distributed ED in a 118 bus system [83] microgrid 2018 + Considered bus power UC/ Centralized Centralized decision-making JADE Demand, renewable, storage constraints Energy transport network in S. Hajjar et al. + Energy storage considered algorithm France (RTE) [87] + reduced communication overhead UC/Hierarchical Simple negotiation strategies Java Demand, Reserve, Ramp-limit Simulation on 10 unit system T. Nagata et al. + obtained results close to dynamic constraints [31] programming dynamic programming and rule- Java Power and reserve limit, system Simulation on 10 unit system J. Yu et al. [88] + agents can solve complex optimization based method constraints 2005 problems Cooperative co-evolution algorithm Mathematica PHEV limit, 10-unit system with standard X. Zhang et al. + considered uncertainty associated with Generation constraints, Ramp rate limits input data of power plants [89] wind, solar and load 2016 + Considered PHEVs UC/ Distributed Zonal optimization approach MATLAB Demand constraints Realistic grid with two E. Kaegi et al. + Two-stage optimization model distributed energy sources [90] + There is no leader node in the model Real-parametric genetic algorithm JADE Demand, ramp limit, reserve constraints Simulation on a 10 unit system D. Sharma et al. + utilized genetic algorithm to solve the (GA) [91] optimization problem 2011 + Rule based intelligence added to ISO agents Technol Econ Smart Grids Sustain Energy (2018) 3:15 Page 11 of 15 15 Fig. 3 Smart grid communicating agents in a centralized UC [87] satisfy its internal demand as the first step, the second zone to reach the profit maximization. In the next stage, stage being contacting the network to analyze the possi- the optimization happens during the interaction between bility of exporting power, and the final step is to schedule zones. The paper only focused on the intra zonal activ- the whole microgrid considering both internal demand ities, but the inter-zonal activities also play a significant and the power transfer from the second stage. The authors role in profit maximization. Figure 5 represents the zon- al approach used by the authors. A summary of the used the JADE platform to simulate the MAS system and used Lagrangian Relaxation with Genetic Algorithm to work done by different researchers on ED and UC prob- schedule the microgrid resources internally. They report lem which utilized MAS concepts are summarized in the robustness and scalability of the method by testing it the Table 3. in a PoolCo energy market. E. Kaegietal. [90] proposed a decentralized ap- proach to solve the UC problem using the MAS con- Conclusion cept. The methodology was based on zonal approach consisting of generator agents, load agents and zone This paper presented a literature review on the applica- agents. The generator agents (GA) and load agents tion of MAS for ED and UC in a smart grid. The inte- (LA) handle the local profit maximization within a zone gration of DERs into a grid requires a decentralized con- while the zone agent handles the interaction with other trol strategy to incorporate these resources and to main- zones. The zone agents have no financial objectives but tain the grid resiliency. The multi-agent technology is a only acts as a service agent for the entities within its promising and scalable platform to implement distributed zone. The optimization is done in two stages. The intra resource scheduling and allocation using various compu- zone level is the competition between agents within a tational techniques. Fig. 4 MAS based UC model [31] 15 Page 12 of 15 Technol Econ Smart Grids Sustain Energy (2018) 3:15 Fig. 5 Agent hierarchy in the UC Model [90] Though there are many centralized algorithms being model which considers the increased DER penetration into the used to solve the ED problem, a small change in the smart future grid. Most of the reviewed articles focused on the im- grid may lead to redesign of these centralized approaches. plementation and convergence ability of the proposed Thus, there is a need for a distributed ED approach which methods, more work needs to be done on evaluating these can enjoy the benefit of robustness, scalability and less methods for their speed and cost savings in a real -time information requirement. Different distributed algorithms environment. for solving ED problem have been proposed by many We believe that this paper can act as a resource for re- researchers in the literature. Of all these distributed ap- searchers in academia and analysts in utilities to understand proaches, consensus-based algorithm has evolved as the the background on MAS’s application for smart grid manage- promising computing method for solving ED. The ment and control. consensus-based ED algorithm can make the analysis Acknowledgments The authors acknowledge the support of the National tractable by simplifying the system into linear for the it- Science Foundation (NSF) award #1537565 for this work. eration process. Most of the consensus-based algorithms available in the literature are useful in solving only con- Open Access This article is distributed under the terms of the Creative vex ED problem without transmission losses. On the other Commons Attribution 4.0 International License (http:// hand, an auction-based algorithm has been proposed to creativecommons.org/licenses/by/4.0/), which permits unrestricted use, solve nonconvex ED problem. However, most of the in- distribution, and reproduction in any medium, provided you give appro- vestigations reported in the literature are limited to imple- priate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. mentation in the simulation environment without address- ing the challenges of different scenarios of a smart grid in real time. Hence, these approaches have to be established References in real time which would be helpful in solving ED prob- lem in a smart grid. 1. Xu Y, Zhang W, Wenxin L (2015) Distributed Dynamic Researchers have explored the application of MAS in cen- Programming-Based Approach for Economic Dispatch in Smart tralized, hierarchical and distributed models for an UC prob- Grids. IEEE Trans Ind Informatics 11:166–175. https://doi.org/10. lem. Most recent works focused on a distributed model utiliz- 1109/TII.2014.2378691 ing concepts from the negotiation strategies and genetic algo- 2. Logenthiran T, Srinivasan D, Khambadkone AM (2011) Multi- agent system for energy resource scheduling of integrated rithm. The distributed UC models used a zonal approach microgrids in a distributed system. Electr Power Syst Res 81: consisting of controller/zonal agents facilitating the commu- 138–148. https://doi.org/10.1016/j.epsr.2010.07.019 nication among agents and these models are not completely 3. Arriagada E, Lopez E, Lopez M et al (2015) A probabilistic eco- distributed. There is a need for a completely distributed UC nomic dispatch model and methodology considering renewable Technol Econ Smart Grids Sustain Energy (2018) 3:15 Page 13 of 15 15 energy, demand and generator uncertainties. Electr Power Syst Res 22. Tolbert LM, Qi H, Peng FZ (2001) Scalable multi-agent system for 121:325–332. https://doi.org/10.1016/j.epsr.2014.11.018 real-time electric power management. 2001 Power Eng Soc 4. Nair AS, Ranganathan P, Kaabouch N (2017) A constrained topo- Summer Meet Conf Proc (Cat No01CH37262) 3:1–4. https://doi. org/10.1109/PESS.2001.970327 logical decomposition method for the next-generation smart grid. In: 2017 Int. Conf. Electr. Comput. Commun. Technol. (IEEE 23. Zheng G, Li N (2010) Multi-Agent Based Control System for ICECCT 2017). Coimbatore, pp 1–6. https://doi.org/10.1109/ Multi-Microgrids. 2010 Int Conf Comput Intell Softw Eng. ICECCT.2017.8118053 https://doi.org/10.1109/CISE.2010.5676818 5. Sujil A, Verma J, Kumar R (2016) Multi agent system: concepts, 24. Wang H (2001) Multi-agent co-ordination for the secondary voltage platforms and applications in power systems. Artif Intell Rev 1–30. control in power-system contingencies. IEE Proc - Gener Transm https://doi.org/10.1007/s10462-016-9520-8 Distrib 148:61–66. https://doi.org/10.1049/ip-gtd:20010025 6. Wooldridge M (2009) An introduction to multiagent systems. 25. Xu Y, Liu W, Gong J (2011) Stable multi-agent-based load shed- https://www.cs.ox.ac.uk/people/michael.wooldridge/pubs/imas/ ding algorithm for power systems. IEEE Trans Power Syst 26: IMAS2e.html. Accessed 5 Jan 2018 2006–2014. https://doi.org/10.1109/TPWRS.2011.2120631 7. Cai N, Nga NTT, Mitra J (2012) Economic dispatch in microgrids 26. Eddy YSF, Gooi HB, Chen SX (2015) Multi-agent system for dis- using multi-agent system. 2012 North Am Power Symp NAPS tributed management of microgrids. IEEE Trans Power Syst 30:24– 2012. https://doi.org/10.1109/NAPS.2012.6336435 34. https://doi.org/10.1109/TPWRS.2014.2322622 8. Bou Ghosn S, Ranganathan P, Salem S, et al (2010) Agent-oriented 27. Abedini R, Pinto T, Morais H, Vale Z (2013) Multi-agent approach designs for a self healing smart grid. 2010 First IEEE Int Conf for power system in a smart grid protection context. 2013 IEEE Smart Grid Commun. https://doi.org/10.1109/SMARTGRID. Grenoble Conf PowerTech, POWERTECH 2013. https://doi.org/ 2010.5622085 10.1109/PTC.2013.6652158 9. Leeton U, Kulworawanichpong T (2012) Multi-Agent Based 28. Wan H, Wong KP, Chung CY (2008) Multi-agent application in Optimal Power Flow Solution. In: 2012 Asia-Pacific Power protection coordination of power system with distributed genera- Energy Eng. Conf. pp 1–4. https://doi.org/10.1109/APPEEC. tions. IEEE Power Energy Soc 2008 Gen Meet Convers Deliv 2012.6307223 Electr Energy 21st Century, PES. https://doi.org/10.1109/PES. 10. Mohammadi J, Hug G, Kar S (2016) Agent-based distributed secu- 2008.4596261 rity constrained optimal power flow. IEEE Trans Smart Grid 9: 29. Do Nascimento LL, Rolim JG (2013) Multi-agent system for adap- 1118–1130. https://doi.org/10.1109/TSG.2016.2577684 tive protection in microgrids. 2013 IEEE PES Conf Innov Smart 11. Ren F, Zhang M, Soetanto D, Su X (2012) Conceptual design of a Grid Technol ISGT LA 2013. https://doi.org/10.1109/ISGT-LA. multi-agent system for interconnected power systems restoration. 2013.6554435 IEEE Trans Power Syst 27:732–740. https://doi.org/10.1109/ 30. Yong L, Tao S (2007) Economic dispatch of power system incor- TPWRS.2011.2177866 porating wind power plant. 2007 Int Power Eng Conf (IPEC 2007) 12. Nagata T, Sasaki H (2002) A multi-agent approach to power system 159–162. https://ieeexplore.ieee.org/document/4510019. Accessed restoration. IEEE Trans Power Syst 17:457–462. https://doi.org/10. 5 Jan 2018 1109/TPWRS.2002.1007918 31. Nagata T, Ohno M, Sasaki H, Fujita H (2002) A multi-agent ap- 13. Solanki JM, Khushalani S, Schulz NN (2007) A multi-agent solu- proach to unit commitment. Electr Eng Japan 141:41–47. https:// tion to distribution systems restoration. IEEE Trans Power Syst 22: doi.org/10.1002/eej.10057 1026–1034. https://doi.org/10.1109/TPWRS.2007.901280 32. Li N, Singhal NG, Hedman KW (2017) An enhanced security- 14. Khamphanchai W, Pisanupoj S, Ongsakul W, Pipattanasomporn M constrained unit commitment model with reserve response set pol- (2011) A multi-agent based power system restoration approach in icies. Proc. 50th Hawaii Int Conf Syst Sci. https://doi.org/10.24251/ distributed smart grid. Util Exhib Power Energy Syst Issues hicss.2017.371 Prospect Asia (ICUE), 2011 Int Conf. https://doi.org/10.1109/ ICUEPES.2011.6497754 33. Kantamneni A, Brown LE, Parker G, Weaver WW (2015) Survey 15. Solanki JM, Schulz NN, Gao W (2005) Reconfiguration for resto- of multi-agent systems for microgrid control. Eng Appl Artif Intell ration of power systems using a multi-agent system. Power Symp 45:192–203. https://doi.org/10.1016/j.engappai.2015.07.005 2005 Proc 37th Annu North Am. https://doi.org/10.1109/NAPS. 34. Dimeas AL, Hatziargyriou ND (2004) A multiagent system for 2005.1560571 microgrids. IEEE Power Eng Soc Gen Meet 2:55–58. https://doi. 16. Khamphanchai W, Pipattanasomporn M, Rahman S (2012) A org/10.1109/PES.2004.1372752 multi-agent system for restoration of an electric power distribution 35. Wang Z, Yang R, Wang L (2011) Intelligent multi-agent control for network with local generation. IEEE Power Energy Soc Gen Meet. integrated building and micro-grid systems. IEEE PES Innov Smart https://doi.org/10.1109/PESGM.2012.6345106 Grid Technol Conf Eur ISGT Eur. https://doi.org/10.1109/ISGT. 17. Raju L, Morais AA, Rathnakumar R et al (2017) Micro-grid Grid 2011.5759134 Outage Management using Multi Agent Systems. Energy Procedia 36. Lü P, Zhao J, Yao J, Yang S (2017) A decentralized approach for 117:112–119. https://doi.org/10.1016/j.egypro.2017.05.113 frequency control and economic dispatch in smart grids. IEEE J 18. Trigo P, Marques P (2008) The electricity market as a multi-agent Emerg Sel Top Circuits Syst 7:447–458. https://doi.org/10.1109/ system. 2008 5th Int Conf Eur Electr Mark EEM. https://doi.org/10. JETCAS.2017.2708900 1109/EEM.2008.4579092 37. Nygard KE, Ghosn SB, Mcculloch R, et al (2012) Decision support 19. Logenthiran T, Srinivasan D (2011) Multi-agent system for market independence in a smart grid. ENERGY 2012 Second Int Conf based microgrid operation in smart grid environment. Dept. Elec. Smart Grids, Green Commun IT Energy-aware Technol Decis Comp. Sci. National University of Singapore 38. Silvestre D, Hespanha JP, Silvestre C (2018) Broadcast and gossip 20. Kok K (2010) Multi-agent coordination in the electricity grid, from stochastic average consensus algorithms in directed topologies. concept towards market introduction. Proc 9th Int Conf Auton IEEE Trans Control Netw Syst. https://doi.org/10.1109/TCNS. Agents Multiagent Syst Ind track 2018.2839341 21. Santos G, Pinto T, Morais H et al (2015) Multi-agent simulation of 39. Raffaelli G, Marsili M (2005) Statistical mechanics model for the competitive electricity markets: Autonomous systems cooperation emergence of consensus. Phys Rev E Stat Nonlinear Soft Matter for European market modeling. Energy Convers Manag 99:387– Phys 72. https://doi.org/10.1103/PhysRevE.72.016114 399. https://doi.org/10.1016/j.enconman.2015.04.042 15 Page 14 of 15 Technol Econ Smart Grids Sustain Energy (2018) 3:15 40. Gu D, Wang Z (2009) Leader-follower flocking: Algorithms and 60. Yang S, Tan S, Xu J-X (2013) Consensus Based Approach for experiments. IEEE Trans Control Syst Technol 17:1211–1219. Economic Dispatch Problem in a Smart Grid. IEEE Trans Power https://doi.org/10.1109/TCST.2008.2009461 Syst 28:4416–4426. https://doi.org/10.1109/TPWRS.2013. 41. Briñón-Arranz L, Schenato L (2013) Consensus-based source- seeking with a circular formation of agents. In: Eur. Control Conf. 61. Binetti G, Naso D, Turchiano B, et al (2014) Consensus-based pp 2831–2836. https://doi.org/10.23919/ECC.2013.6669782 approach for the economic dispatch problem. In: Proc. IFAC 42. Zhang X, Xu H, Yu T et al (2016) Robust collaborative consensus World Congr. IFAC, pp 3140–3145. https://doi.org/10.3182/ algorithm for decentralized economic dispatch with a practical com- 20140824-6-ZA-1003.01786 munication network. Electr Power Syst Res 140:597–610. https:// 62. Cherukuri A, Cortés J (2014) Distributed coordination for econom- doi.org/10.1016/j.epsr.2016.05.014 ic dispatch under varying loads and generator commitment. Fifty- 43. Hug G, Kar S, Wu C (2015) Consensus+Innovations Approach for second Annu Allert Conf. https://doi.org/10.1109/allerton.2014. Distributed Multiagent Coordination in a Microgrid. IEEE Trans Smart Grid 6:1893–1903. https://doi.org/10.1109/TSG.2015. 63. Luo K, Wu Q, Jacob Ø, Hejde NA (2015) Real-time distributed economic dispatch for distributed generation based on multi-agent 44. Lü P, Zhao J, Yao J, Yang S (2017) A decentralized approach for system. 2015 Mod. Electr. Power Syst. https://doi.org/10.1109/ frequency control and economic dispatch in smart grids. IEEE J MEPS.2015.7477174 Emerg Sel Top Circuits Syst PP:1–12 64. Vergara PP, Shaker HR, Norregaard B, da Silva LCP (2017) 45. Pipattanasomporn M, Feroze H, Rahman S (2009) Multi-agent sys- Distributed consensus-based economic dispatch considering grid tems in a distributed smart grid: Design and implementation. operation. 2017 IEEE Power Energy Soc Gen Meet. https://doi. IEEE/PES Power Syst Conf Expo 2009 PSCE ‘09 1–8. https:// org/10.1109/PESGM.2017.8273864 doi.org/10.1109/PSCE.2009.4840087 65. Guo F, Wen C, Mao J, Song YD (2016) Distributed Economic 46. Kravari K, Bassiliades N (2015) A survey of agent platforms. Dispatch for Smart Grids with Random Wind Power. IEEE Trans JASSS. https://doi.org/10.18564/jasss.2661 Smart Grid 7:1572–1583. https://doi.org/10.1109/TSG.2015.2434831 47. Singh A, Juneja D, Sharma AK (2011) Agent Development 66. Cao J, Yu M, Tung LJ (2014) Consensus-based distributed control Toolkits. Int J Adv Technol 2:158–164 for economic dispatch problem with comprehensive constraints in a 48. Shoham Y (1991) AGENT0: A simple agent language and its in- smart grid. EAI Endorsed Trans Energy Web 1:1–9. https://doi.org/ terpreter. In: Proc. AAAI-91. pp 704–709 10.4108/ew.1.2.e3 49. Thomas SR (1994) The PLACA agent programming language. In: 67. Yang Z, Xiang J, Li Y (2016) Distributed virtual incremental cost Intell. Agents. pp 355–370 consensus algorithm for economic dispatch in a microgrid. IEEE Int 50. Bellifemine F, Caire G, Greenwood D (2007) Developing multi- Conf Control Autom ICCA. https://doi.org/10.1109/ICCA.2016. agent systems with JADE. John Wiley Sons Ltd. https://doi.org/10. 1002/9780470058411 68. Li Y-S, Zhang H-G, Huang B-N, Teng F (2016) Distributed optimal 51. Gutknecht O, Ferber J (2001) The MadKit agent platform architec- economic dispatch based on multi-agent system framework in com- ture. In: Infrastruct. Agents, Multi-Agent Syst. Scalable Multi- bined heat and power systems. Appl Sci. https://doi.org/10.3390/ Agent Syst. SE - 5. pp 48–55 app6100308 52. Guo Q, Han J, Yoon M, Jang G (2012) A study of economic dis- 69. Yang Z, Xiang J, Li Y (2017) Distributed consensus based supply- patch with emission constraint in smart grid including wind turbines demand balance algorithm for economic dispatch problem in a and electric vehicles. 2012 IEEE Veh Power Propuls Conf VPPC smart grid with switching graph. IEEE Trans Ind Electron 64: 2012 1002–1005. https://doi.org/10.1109/VPPC.2012.6422753 1600–1610. https://doi.org/10.1109/TIE.2016.2615037 53. Nair AS, Ranganathan P, Salehfar H, Kaabouch N (2017) 70. Xing H, Mou Y, Fu M, Lin Z (2014) Consensus based bisection Uncertainty quantification of wind penetration and integration into approach for economic power dispatch. Proc IEEE Conf Decis smart grid: a survey. North Am. Power Symp. https://doi.org/10. Control 2015–Febru. https://doi.org/10.1109/CDC.2014.7039979 1109/NAPS.2017.8107196 71. Binetti G, Davoudi A, Naso D, et al (2014) A distributed auction- 54. Dominguez-Garcia AD, Cady ST, Hadjicostis CN (2012) based algorithm for the nonconvex economic dispatch problem. In: Decentralized optimal dispatch of distributed energy resources. IEEE Trans. Ind. Informatics 10:1124–1132. https://doi.org/10. In: Proc. IEEE Conf. Decis. Control. pp 3688–3693. https://doi. 1109/TII.2013.2287807 org/10.1109/CDC.2012.6426665 72. Binetti G, Davoudi A, Lewis FL et al (2014) Distributed consensus- 55. Rana R, Bhattacharjee S, Mishra S A consensus based solution to based economic dispatch with transmission losses. IEEE Trans the classic economic dispatch problem on a multi-agent system Power Syst 29:1711–1720. https://doi.org/10.1109/TPWRS.2014. framework. In: 2018 IEEMA Eng. Infin. Conf. IEEE, pp 1–6. https://doi.org/10.1109/ETECHNXT.2018.8385372 73. Zhabelova G, Vyatkin V, Zhang Z, Chow MY (2013) Agent-based 56. Yu W, Li C, Yu X et al (2018) Economic power dispatch in smart distributed consensus algorithm for decentralized economic dis- grids: a framework for distributed optimization and consensus dy- patch in Smart Grid. In: IECON Proc. (Industrial Electron. Conf. namics. Sci China Inf Sci 61:1–16. https://doi.org/10.1007/s11432- Vienna, Austria. https://doi.org/10.1109/IECON.2013.6699433 016-9114-y 74. Wu C, Li H, Wu L, Wu Z (2015) A Multi-Agent Particle Swarm 57. Zhang Z, Chow MY (2011) Incremental cost consensus algorithm Optimization for Power System Economic Load Dispatch. J in a smart grid environment. IEEE Power Energy Soc Gen Meet. Comput Commun 3:83–89 https://doi.org/10.1109/PES.2011.6039422 75. Kumar R, Sharma D, Sadu A (2011) A hybrid multi-agent based 58. Zhang Z, Chow M (2012) Convergence Analysis of the Incremental particle swarm optimization algorithm for economic power dis- Cost Consensus Algorithm Under Different Communication patch. Int J Electr Power Energy Syst 33:115–123. https://doi.org/ Network Topologies in a Smart Grid. IEEE Trans Power Syst 27: 10.1016/j.ijepes.2010.06.021 1761–1768 59. Zhang Z, Ying X, Chow MY (2011) Decentralizing the economic 76. Loia V, Vaccaro A (2014) Decentralized economic dispatch in smart dispatch problem using a two-level incremental cost consensus al- grids by self-organizing dynamic agents. IEEE Trans Syst Man, gorithm in a smart grid environment. NAPS 2011 - 43rd North Am Cybern Syst 44:397–408. https://doi.org/10.1109/TSMC.2013. Power Symp. https://doi.org/10.1109/NAPS.2011.6025103 2258909 Technol Econ Smart Grids Sustain Energy (2018) 3:15 Page 15 of 15 15 77. Zhao C, He J, Cheng P, Chen J (2017) Analysis of Consensus- 88. Yu J, Zhou J, Hua B, Liao R (2005) Optimal Short-Term Generation Based Distributed Economic Dispatch Under Stealthy Attacks. Scheduling with Multi-Agent System under a Deregulated Power. IEEE Trans Ind Electron 64:5107–5117 Int J Comput Cogn 3:61–65 89. Zhang X, Xie J, Zhu Z et al (2016) Smart grid cost-emission unit 78. Li C, Yu X, Yu W et al (2016) Distributed event-triggered scheme commitment via Co-evolutionary agents. Energies 9:1–13. https:// for economic dispatch in smart grids. IEEE Trans Ind Informatics doi.org/10.3390/en9100834 12:1775–1785. https://doi.org/10.1109/TII.2015.2479558 90. Kaegi E, Cherkaoui R, Germond A (2008) Decentralized unit com- 79. Zhang Y, Rahbari-Asr N, Chow MY (2016) A robust distributed mitment and dispatch for the distribution systems using intelligent system incremental cost estimation algorithm for smart grid econom- agents approach. In: PSCC 2008 ic dispatch with communications information losses. J Netw Comput 91. Sharma D, Srinivasan D, Trivedi A (2011) Multi-agent approach for Appl 59:315–324. https://doi.org/10.1016/j.jnca.2015.05.014 profit based unit commitment. In: 2011 IEEE Congr. Evol. Comput. 80. Wen G, Yu X, Liu ZW, Yu W (2017) Adaptive Consensus-Based CEC 2011. https://doi.org/10.1109/CEC.2011.5949932 Robust Strategy for Economic Dispatch of Smart Grids Subject to 92. Padhy NP (2004) Unit Commitment—A Bibliographical Survey. Communication Uncertainties. IEEE Trans Ind Informatics 14: IEEE Trans Power Syst 19:1196–1205. https://doi.org/10.1109/ 2484–2496. https://doi.org/10.1109/TII.2017.2772088 TPWRS.2003.821611 81. Hamdi M, Chaoui M, Idoumghar L, Kachouri A (2018) 93. Carrión M, Arroyo JM (2006) A computationally efficient mixed- Coordinated consensus for smart grid economic environmental integer linear formulation for the thermal unit commitment prob- power dispatch with dynamic communication network. IET lem. IEEE Trans Power Syst 21:1371–1378. https://doi.org/10. Gener Transm Distrib 12:2603–2613. https://doi.org/10.1049/iet- 1109/TPWRS.2006.876672 gtd.2017.1197 94. Lopez CJ, Ano O, Ojeda Esteybar D (2018) Stochastic Unit 82. Yang T, Lu J, Wu D et al (2017) A Distributed Algorithm for Commitment & Optimal Allocation of Reserves: A Hybrid Economic Dispatch over Time-Varying Directed Networks with Decomposition Approach. IEEE Trans Power Syst 8950:1–11. Delays. IEEE Trans Ind Electron 64:5095–5106. https://doi.org/ https://doi.org/10.1109/TPWRS.2018.2817639 10.1109/TIE.2016.2617832 95. Wang Y, Shahidehpour M, Lai LL et al (2018) Resilience- 83. Wang R, Li Q, Zhang B, Wang L (2018) Distributed Consensus Based Constrained Hourly Unit Commitment in Electricity Grids. IEEE Algorithm for Economic Dispatch in a Microgrid. IEEE Trans Smart Trans Power Syst 8950:1–10. https://doi.org/10.1109/TPWRS. Grid 0:1–11. https://doi.org/10.1109/TSG.2018.2833108 2018.2817929 84. Tang Z, Hill DJ, Liu T (2018) A Novel Consensus-Based Economic 96. Xiong P, Singh C (2018) A Distributional Interpretation of Dispatch for Microgrids. IEEE Trans Smart Grid 9:3920–3922. Uncertainty Sets in Unit Commitment under Uncertain Wind https://doi.org/10.1109/TSG.2018.2835657 Power. IEEE Trans Sustain Energy 3029:1–9. https://doi.org/10. 85. He X, Yu J, Huang T, Li C (2018) Distributed power management 1109/TSTE.2018.2828421 for dynamic economic dispatch in the multimicrogrids environ- 97. Abedi S, He M, Member S, et al (2018) Congestion risk-aware unit ment. IEEE Trans Control Syst Technol 1–8. https://doi.org/10. commitment with significant wind power generation. IEEE Trans 1109/TCST.2018.2816902 Power Syst 33:6861–6869. https://doi.org/10.1109/TPWRS.2018. 86. Lin P, Jin C, Xiao J et al (2018) A Distributed Control Architecture for Global System Economic Operation in Autonomous Hybrid 98. Nagata T, Ohono M, Kubokawa J et al (2002) A multi-agent approach AC/DC Microgrids. IEEE Trans Smart Grid 3053:1–13. https:// to unit commitment problems. Proc IEEE Power Eng Soc Transm doi.org/10.1109/TSG.2018.2805839 Distrib Conf 1:64–69. https://doi.org/10.1109/PESW.2002.984955 87. Hajjar S, Bratcu AI, Hably A (2015) A day-ahead centralized unit 99. Yu J, Zhou J, Wu W, et al (2004) Solution of the profit-based unit commitment algorithm for a multi-agent smart grid. In: Fed Conf commitment problem by using multi-agent system. In: WCICA Comput Sci Inf Syst. pp 265–271. https://doi.org/10.15439/ 2004. Fifth World Congr. Intell. Control Autom. https://doi.org/ 2015f227 10.1109/WCICA.2004.1343686 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Technology and Economics of Smart Grids and Sustainable Energy Springer Journals

Loading next page...
 
/lp/springer-journals/multi-agent-systems-for-resource-allocation-and-scheduling-in-a-smart-0uhrNu8YyA
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Energy; Energy Systems; Power Electronics, Electrical Machines and Networks; Energy Economics
eISSN
2199-4706
DOI
10.1007/s40866-018-0052-y
Publisher site
See Article on Publisher Site

Abstract

With the increasing integration of Distributed Energy Resources (DER) in the power grid, a decentralized approach becomes essential for scheduling and allocation of resources in a smart grid. Economic Dispatch (ED) and Unit Commitment (UC) are the two major resource allocation problems that play critical role in the safe and stable operation of a grid system. The uncertainty associated with renewable energy sources have made the resource allocation problems even more challenging for grid operators. The future grid will have a higher generation mix of renewable energy sources and a large load of Electrical vehicles, with the possibility of bi-directional power flow. This complex smart grid system necessitates the development of a decentralized approach to resource allocation problem, which allows inter-node communication and decision making. Multi-agent systems (MAS) is a promising platform to decentralize the traditional centralized resource allocation aspects of smart grid. This paper presents a comprehensive literature review on the application of MAS to Economic Dispatch (ED) and Unit Commitment (UC) in smart grids. . . . . Keywords Multi-agent systems Smart grid Resource allocation and scheduling Economic dispatch Unit commitment Introduction The presence of increased penetration of Renewable Energy Sources (RES) in the power system creates many The smart grid framework incorporates distributed generation, technological challenges for the power companies owing to advanced communication network, smart meters and sensors the need for improved system control to maintain the power to make the grid more reliable, flexible, adaptive and efficient. quality to consumers [3]. The mix of conventional and RES This new power system paradigm necessitates the need to must work in tandem to maintain the power generation at the revisit some of the traditional power system operations to meet required level. The process of committing the generators and the challenges of next-generation transmission and distribu- allocating the required generation level has become a chal- tion systems [1]. Along with the integration of renewable en- lenge to meet the increased demand, while using the genera- ergy sources, the deregulation in the energy market has creat- tion from uncertain RES. These factors make the centralized ed competition for power generation companies. Generation control of a smart grid system complex and less efficient to companies have an obligation to meet the customer energy process the diversity of data and controls [4]. The concept of demands even during peak hours and system outages. There Multi-Agent Systems (MAS) is put forward to solve this prob- is a need to properly allocate the generation sources to maxi- lem by using automated agent technology. The MAS converts mize the profit considering renewable generation and custom- a centralized control system into a distributed control model at er demand [2]. a component level. MAS is a collection of agents working together with each other to achieve an overall objective [5]. An agent can be defined as a computer system with the ability to take critical * Arun Sukumaran Nair arun.sukumarannair@und.edu decisions based on the scenario to improve its objective [6, 7]. These software agents in a smart grid environment sense, 1 communicate, collaborate and act with each other. The agents Department of Electrical Engineering, University of North Dakota, can act autonomously or semi-autonomously, with local or Grand Forks, ND, USA 15 Page 2 of 15 Technol Econ Smart Grids Sustain Energy (2018) 3:15 global information [8]. MAS technology is finding wide range The centralized MAS architecture for micro grid control is of applications in the power system domain such as optimal discussed in the literature [34, 35]. The framework of distrib- power flow [9, 10], power system restoration [11–17], elec- uted and three level hierarchical MAS for a smart grid is ex- tricity market operation [18–21], power system control plained inP. Luetal. [36] and K. E. Nygard et al. [37] [22–26] and protection [27–29]. The focus of this paper will respectively. be restricted to the application of MAS in the fundamental Most of the MAS based optimization problems rely on the resource allocation aspects of the power grid namely, technique of consensus algorithm to reach the solution. The Economic Dispatch (ED) and Unit Commitment (UC). main idea behind the consensus problem is to make a set of ED is one of the most important challenges in the pow- agents agree up on a certain value (usually a global function) er systems and it deals with the allocation of power gen- by using local information exchange among agents (local in- eration among committed generators in order to meet the teraction). This concept is utilized in different fields such as demand while lowering the generation cost [30]. economics (Agreement problem), communication (Gossip al- Consumer demand for clean energy and government reg- gorithms [38]), statistical mechanics [39] (Synchronization) ulation has motivated the integration of Distributed and robotics (Flocking [40]). It offers several advantages over Energy Resources (DERs) like solar photovoltaic, wind traditional centralized methods such as distributed computa- power, and fuel cells into the modern power grid. This tion, computational efficiency, independent of graph topology makes ED a highly complex optimization problem which and robustnesstofailure [41, 42]. In a consensus algorithm needs to consider the various factors like generator capac- model, each node in the system is considered as a dynamic ity, ramp-rate, failure rate, emission, load profile and gen- agent with a value or state associated with it. The value of the eration from DER. Unit Commitment (UC) in a smart grid agent represents the decision variable with which it can reach system is a highly complex optimization problem that consensus with other agents in the system. Researchers have schedules the startup and shutdown of generators to meet explored different census methods for a microgrid system the demand while satisfying system constraints [31, 32]. such as in the work by G. Hug et al. [43] where a combination The committed generators are modeled in the ED for gen- of consensus and innovation method was utilized. A novel erator scheduling. The smart grid systems which have framework to model a full automation of a distributed smart significant DER and the increased interest from con- grid system is presented in the work by K. E. Nygard et al. sumers to install RES have necessitated the need for a [37]. The model is based on the concept of an Intelligent decentralized approach to commit and schedule genera- Autonomous Distributed Power System (IDAPS), a microgrid tors. The increased uncertainty from RES has made ED with sufficient resources and intelligence to function autono- and UC more complicated due to the intermittent nature mously within a global grid. A three-layer hierarchical system of these power sources. model with agents in higher level supervising the agents in For understanding the application of MAS in resolving lower levels is proposed. The model accomplishes modularity, resource allocation and scheduling problems of smart grid, scalability, and a balance between global and local decisions the paper is organized in five sections. Section II describes of agents. Distributed MAS based control offers several ad- the architecture of a MAS in a smart grid system, section III vantages such as autonomy, fault-tolerance low latency, effi- and IV present comprehensive review on the application of ciency and much more. It is a way of physically breaking MAS in ED and UC respectively, followed by conclusion in complex control problems into smaller control problems, section V. and then solving them closer to the control operation itself. Development of Smart Grids will involve dealing with a bigamountofdatacollectedinadistributedmanner.Thisdata is communicated among equipment and devices to support MAS for Smart Grid decision-making process. Certainly, handling the amounts of data to be acquired and processed in such distributed systems An agent represents a computer system situated in an environ- to extract useful information bring its own challenges. ment where it is capable of making decisions to achieve its Computational intelligence techniques are used to extract design objectives. Moreover, an agent can be autonomous, knowledge and overcome some of the challenges. In “intelli- social, reactive and proactive. Multi Agent Systems (MAS) gent” systems, data is pre-processed, processed, and then are composed of agents interacting in a highly dynamic envi- information is extracted for decision-making. Given the dis- ronment. These intelligent agents are being developed to have tributed nature of Smart Grids (SG), new advancements in the functionalities on par with the human experts to act appro- distributed intelligence techniques spawned the MAS tech- priately in the various scenarios that take place in a smart grid. nique development. The following are some advantages and The summary of various MAS architectures used for control reasons to explore distributed MAS architectures over cen- tralized architecture: of microgrid is summarized in Table 1 [33]. Technol Econ Smart Grids Sustain Energy (2018) 3:15 Page 3 of 15 15 Table 1 Different MAS architectures for microgrid control [33] MAS architecture Type of agent Role Features Centralized Cognitive Agent Higher level of intelligence/ + collects information communication capabilities at asinglepoint Reactive Agent Fast Response + capable of making global decision + flexibility and openness in the operation of smart grid − suffers from computational burden in case of large number of agents − single point of failure affects the entire system Two-level hierarchical High level agent Infrastructure management, + distinct levels of decision making low level scheduling − failure of higher level agents results Low level agent Accept schedule from High in critical conditions of the lower level agents level agent, asset management Three-level hierarchical High level agent Critical decisions, data and + good scalability through delineation policy management of roles to agents Middle level agent Fault location, switching of grid − failure of higher level agents results in connected/islanded mode critical conditions of the lower level agents Low level agent Sensor management Distributed Local Agent Local information discovery/ + robust system with agents being capable communication of reorganizing and coping up with the loss of other agents Advantages of Distributed MAS over centralized Response (DR) with the ability to dynamically optimize control: grid operations , resources and consumer participation. To do so, it is important to understand demand participation 1. The SG components are often distributed and the of consumers. As number of consumers are growing, it is energy management system is tightly associated with essential to do the demand response analytics in a distrib- the communications between stakeholders and enti- uted fashion. ties (agents) to exchange information, so MAS is an appropriate platform to develop distributed manage- Due to the inherent advantages of distributed MAS archi- ment functions. tecture, it is well suited to resolve the complex ED problem of 2. SG is a holistic system and the failure of some part of it smart grid. A general framework for the implementation of (e.g., the breakdown of a transmission line or cut down of MAS in smart grid (SG) system is shown in Fig. 1 [44]. SG a substation, transformer) should not affect the whole ac- system is an integration of the physical grid with the commu- tivities and operations, and hence fault tolerance can be nication layer where the agents act as an interface. The com- easily attained in distributed architecture over centralized munication layer is a strongly connected network with varying architecture. and configurable topologies. Each agent can be categorized 3. SG should demonstrate the plug-and-play concept for in- into three units; namely Device Unit (DU), Decision Making tegrating energy storage, loads, and sources at the build- Unit (DMU) and Communication Unit (CU). DUs can be ing level with the external utility grid. Plug and play considered as physical power system buses with components adaptability is widely proven by MAS. The nature of such as Synchronous Generators (SGR), Renewable MAS enables it to scale up by adding other agents or by Generators (RG), flexible load and rigid load. DMUs perform dispersing them in new environment with new resources the local computing for the agents and CUs are the communi- and capacities. Hence, Distributed MAS building mod- cation nodes, which transmit and receive information [45]. ules are highly scalable, and modular. The internal structure of an agent model is shown in Fig. 2. 4. As SG will be composed of an aggregate of Micro grid, An agent model consists of three units; Communication Unit and hence the control can be delegated to micro grids. (CU), Decision Making Unit (DMU) and Device Unit (DU). With futuristic smart grids being a simple collection of CUs are generally signal receivers/transmitters used to ex- residential microgrids, each microgrid can exhibit distrib- change information with neighbors. The calculator, sensors uted control. and controllers are part of DMU, which are responsible for 5. One of the goals of the SG is to develop grid moderniza- the local computing in an agent. DMU is the brain of an agent tion technologies, tools, and techniques for Demand- node and capable of generating control instructions for the DU 15 Page 4 of 15 Technol Econ Smart Grids Sustain Energy (2018) 3:15 Fig. 1 Framework for distributed multi-agent system for smart grid Decentralized Communication Network [44] CU CU CU DMU DMU Agents DMU SGR RG Load Physical Grid as well as responsible for communicating the information to generators in the system to meet the demand. ED needs to the CU. DUs represent the traditional buses in a network conform to several other constraints for a safe and secure which consists of different elements such as synchronous gen- operation of the grid. The integration of uncertain renewable erators, Renewable Generators (RG), battery storage systems energy sources to the grid has made ED and power quality (BESSs), flexible and rigid loads. DU performs the control analysis more important and also more complicated [52, 53]. suggestions from DMU and also sends the feedback to the Distributed algorithms are becoming popular for intelligent DMU. decision-making and control and these algorithms appear to There are number of simulation and open source tools be promising in the context of smart grid. These algorithms available for modeling MAS platforms [46]. The most com- are robust, immune to topological variations and can support mon ones are ZEUS, AgentBuilder, JADE, and MADKit. the “plug-and-play” feature of the future grid. However, it is Features of these MAS modeling tools are listed in Table 2. more challenging to include the operational constraints in such a distributed formulation. Many researchers have proposed a consensus-based approach for ED without losses and lower Economic Dispatch and upper power boundaries. A consensus algorithm is widely used in solving the ED problem in smart grid. It is a method Economic Dispatch (ED) is one of the fundamental problems used to achieve agreement on a single data value among dis- in the power system domain. It is basically an optimization tributed systems. This algorithm is designed to achieve reli- problem with the objective of reducing cost while maintaining ability in a network involving multiple unreliable nodes. [36, the generation-load balance. ED schedules the committed 54–56]. Fig. 2 Internal structure of an Agent i agent [44] in out Information Communication Unit (CU) Exchanging clock, signal receiver & transmitter External Internal Information Information Local Decision Making Unit (DMU) Computing monitor, sensor, calculator & controller Sampling Control signals suggestions Device Unit (DU) Power synchronous generators/RG/BESS/load Flow Technol Econ Smart Grids Sustain Energy (2018) 3:15 Page 5 of 15 15 Table 2 MAS modeling tools MAS tools Description ZEUS ZEUS [47] is a multi-agent platform developed by the research program of British Telecom intelligent system research laboratory. ZEUS allows the design of multi-agent distributed systems. This platform, developed in Java, automatically generates Java code from the agents specified graphically. Agent Builder Developed by Reticular Systems Inc., AgentBuilder [47], it is based on BDI (Believe - Desire - Intention) models Agent [48] and AGENT-O language [49]. It is remarkable for the quality of its software and a good academic model. AgentBuilder is a commercial design software for “intelligent” agents i.e., cognitive and collaborative agents. AgentBuilder consists of two main components: the toolkit and runtime system. JADE JADE [50] is a multi-agent (multi-host) platform developed by Bellifemine. F., Poggy. A., Rimassa. G. and P. Turci by Telecom Italia Lab “Tilab formerly CSELT” in 1999. This platform aims to simplify the construction of interoperable MAS, achieve compliant applications with the standard FIP A97 (Foundation for Intelligent Physical Agents) to facilitate the communication of JADE agents with non-JADE agents, and optimize the performance of a distributed system agent. JADE includes all accredited component that manages the platform: Agent Communication Channel (ACC), Agent Management System (AMS), and Director Facilitator (DF). MADKit MADKit [51] is a platform for MAS developed by Olivier Gutknecht and Jacques Ferber in Laboratory of Computer Science and Robotics and Microelectronics of Montpellier. MADKit was motivated by the need for a more flexible platform possible, and able to adapt to different agent models and application areas. MADKit is a modular multi-agent platform and scalable, written in the Java language. It allows the creation of MAS based on the relational model Aalaadin or AGR (Agent / Group / Role): agents are located in groups and play roles. MADKit takes advantage of object-oriented programming: MADKit features are contained in the MADKit kernel. In a conventional centralized method (e.g. Lagrange mul- to acquire system power mismatch. This new approach will tiplier method), at the optimal point, all the generators will eliminate the need for a single leader node to do all the calcu- have the same incremental cost. An appropriate consensus lation. An average consensus will run at the lower level of the algorithm can guarantee a similar result by having all the two-level method and ICC will be employed at the second consensus variables to converge to a common value asymp- level to process the mismatch information. This is an im- totically. Based on this concept, Z. Zhang and M. Chow [57] proved version from the method proposed in [57]. It is more introduced an Incremental Consensus Algorithm (ICC) to de- distributed and does not require a fixed communication centralize the ED problem by choosing incremental cost (IC) network. as a consensus variable. The model consists of a local control- A decentralized approach to ED in a microgrid with ler (generation unit) which will update its consensus variable Distributed Generators (DG) was explored by N. Cai et al. depending up on the neighbor’s values. The proposed ap- [7] using a MAS. Here, each DG was assumed to have an proach requires a leader node, which will decide whether to agent which could receive local information and communicate increase or decrease the IC based on the demand constraints. only with its nearest neighbors. Agents compete with others to The authors tested the approach on a 3 unit and 5 unit system obtain a local solution, thereby obtaining a global optimum. to test the validity and convergence of the proposed approach. The authors used the concept of consensus among agents to They showed a successful implementation of consensus algo- obtain the optimum solution for the ED. The authors validated rithm in ED but lacked a fully distributed model since it need- the approach on five and fifty agent systems but did not com- ed a leader node to control the agents. A more detailed de- pare the results with a centralized approach. scription about ICC is provided in [58]. A consensus control based approach to solve the ED prob- The previous paper utilized ICC algorithm to implement lem in a smart grid was developed by S. Yang et al. [60]. The ED in a distributed fashion but relied on a leader-follower approach solves the ED in a distributed fashion with the gen- consensus algorithm. A leader node needs to be selected, erators acting collectively to receive the mismatch between which will gather the local power mismatch from the follower demand and power generation information, which is the feed- nodes to calculate the total power mismatch. The follower back for the agents. The total mismatch is generated in a col- nodes need to report their power mismatch to their leader. Z. lective fashion from the estimate of local mismatch by the Zhang et al. [59] introduced a two-level consensus approach agents, which removes the need for a leader agent to collect 15 Page 6 of 15 Technol Econ Smart Grids Sustain Energy (2018) 3:15 the global information. The incremental cost of the generators F. Guo et al. [65] explored the potential of a distributed ED is chosen as the consensus variable and incremental cost cri- model for a smart grid system with random wind power. The terion was used to obtain the optimal dispatch. The method proposed model works on the projected gradient and Finite- was found to have the same precision as the Lambda-Iteration time Average Consensus Algorithm (FACA) and supports the approach, a centralized method, with less communication plug-and-play feature of new generation smart grids. The ran- overhead. dom wind power generation is modeled using the determinis- A distributed ED model considering line loss was devel- tic method with overestimation and underestimation cost var- oped by G. Benetti et al. [61]. The nodes in the model run two iables. The agents can choose arbitrary initial values and are consensus methods in parallel; one to find the Lagrangian not required to share gradient or incremental cost information variable and a second one to find the power mismatch. The with the neighbors. The authors validated the effectiveness first method is a first-order consensus algorithm which uses a andperformance of theproposedmethodonIEEEtest proportional controller to bring the power mismatch to zero systems. and to satisfy the generation-demand equality constraint. The A consensus-based distributed ED taking into account gen- second consensus method uses the work allocation concept to erator dynamics was studied by J. Cao et al. in [66]. The find the power mismatch. The authors assert that the proposed authors used comprehensive generator constraints to improve method can satisfy generation constraint and can handle line the consensus algorithm and analyzed the effect of different loss in the system. The comparison between the distributed communication topologies on the speed of the consensus al- approach and the centralized approach to verify its conver- gorithm. The model relies on local power mismatch data from gence speed and accuracy was not attempted by the authors. the agents rather than a leader node to collect global informa- A. Cherukuri et al. [62] explored the concept of distributed tion. The authors asserted the superiority of the proposed consensus-based approach to model an ED which can handle method by comparing with Lambda iteration and PSO changing load conditions and can remain stable under inter- methods. The generator dynamics was found having a signif- mittent power sources. The proposed model employs two dy- icant effect on the speed of the consensus method while the namical systems namely, dynamic average consensus and effect of communication topology was not significant. Laplacian non-smooth gradient. The mismatch between gen- A distributed consensus-based approach to solve ED in a eration and load is estimated in a distributed fashion by the microgrid was developed by Z. Yang et al. [67]. They used a consensus method and the Laplacian non-smooth gradient dy- novel concept of virtual incremental cost as the consensus namically allocates the generation. The approach can reach variable which does not require the nodes to share power optimum solution from any initial power allocation and does output or generator parameters. The algorithm has the advan- not require a feasible allocation as the initial value. The au- tage of not depending on the local power mismatch to reach the optimum and maintaining the supply-demand balance thors verified the effectiveness of the method to handle dy- namic loads and intermittent power sources. even during transients. They reported reduction in communi- K. Luo et al. [63] developed a MAS based distributed ED cational burden between nodes and improved reliability of the model for an electrical grid system with RES, which can be algorithm. deployed for real-time applications. The proposed approach is Y. Li et al. [68] developed a distributed ED model for a a two-step process, with the first step calculating the initial combined heat and power system. The MAS based framework generation values using adjacency average allocation algo- utilized two consensus protocols, one optimizes the electrical rithm and the second stage performs the ED in a distributed incremental cost function while the other gets a common value manner using local replicator dynamics. The first stage han- for the heat incremental cost. The heat and power coupling in dles the equality constraints in the model while the second the objective function and constraints are managed by these stage conforms to the inequality constraints. They validated two consensus variables. It works in a completely distributed the effectiveness of the proposed method but did not compare fashion without the need for a leader agent with the global the performance of the method with similar approaches. information. They report the effectiveness of the proposed A distributed ED model for an islanded microgrid system ED model by comparing to a centralized approach using was developed by P.P. Vergara et al. in [64]. The model con- Lagrangian relaxation method. sidered both active and reactive power in the optimization Z. Yang et al. [69] proposed a distributed consensus-based model. The primal-dual constrained optimization method model for the ED in a smart grid system which maintains the was used to solve the problem in a distributed fashion, in supply-demand balance even during the transient process. The which two consensus methods are executed in parallel to ob- method has the advantage of not relying on the supply-demand tain the dual values or incremental costs. The authors validat- mismatch and hence can be used online. The proposed method ed the performance of the proposed method by comparing to a does not require a leader node with the complete information of classical Lambda method and also the capability of the meth- power demand in the grid system. It uses the maximum incre- od for fault tolerance. mental cost of the neighboring generators and developed a Technol Econ Smart Grids Sustain Energy (2018) 3:15 Page 7 of 15 15 method to increase or decrease the incremental cost of a satu- A dynamic agent-based approach to model a decentralized rated generator to maintain the supply-demand balance during ED was developed by V. Loia et al. [76]. ED was solved using iterations. H. Xing et al. [70] utilized an average consensus self-organizing dynamic agents equipped with distributed based bisection approach to perform distributed ED. The meth- consensus method. C. Zhao et al. [77] explored the effect of od has the advantage of not relying on prior information or a cyber-attacks on a consensus-based ED model. The authors leader node to perform the optimization. tested the performance of the algorithm for false data injection G. Binettietal. [71] developed a distributed model to into the broadcast information, offline and online ED models, solve non-convex ED problems. The non-convexity comes and bounded and unbounded generation cases. from the valve-point effect, prohibited operating zones, The increased amount of communication between nodes in multiple fuel option and transmission losses but makes the a smart grid system can lead to communication bottlenecks modelmorerealisticfor real-timeoperations. Theproposed which can cause convergence issues in consensus-based ED model is fully decentralized and does not require a leader models.C.Lietal. [78] developed an event triggered node with the global information. The method has the added consensus-based ED model to reduce the communication advantage of being deterministic while heuristic methods do overhead in a smart grid system. The authors reformulated not guarantee the uniqueness of the solution from a single the ED model using θ-logarithmic barrier to conduct the in- run. A combination of auction mechanism and market- formation exchange in a distributed fashion. The reformulated based MAS was used to design the distributed ED and the ED model is solved in a two-stage process; the initial values authors tested the validity of the method on standard test for the agents are generated using connected dominating set systems. G. Binetti et al. [72] also proposed a distributed based distribution algorithm as the first stage and in the next ED model which considered transmission losses in the sys- stage a consensus-based optimization is applied to the system. tem using a combination of two consensus algorithms run- The authors stated that asynchronous communication-based ning in parallel. The model utilized a first order consensus event triggered ED model can significantly reduce the com- protocol to calculate the local power mismatch to satisfy the munication exchange in a smart grid system, but the event demand constraint and a second consensus algorithm to cal- triggered mechanisms can have a negative impact on the con- culate the system power mismatch. vergence rate. A fast gradient-based method is used to accel- A transition of the MAS based distributed ED from labo- erate the convergence rate in the optimization model. ratory set up to industrial model is studied by G. Zhabelova Most of the papers discussed above assume a perfect com- et al. [73]. An incremental cost consensus approach model munication between agents without any information loss, but based on the industrial standard IEC 61499 is used to solve in a realistic smart grid environment can have packet loss and ED in a smart grid environment. IEC 61499 is a promising communication failures. Y. Zhang et al. [79] proposed a dis- tributed ED model which remains robust under information industrial standard used as an architecture for the development of distributed systems in control and automation. The agent- loss among agents. A combination of two algorithms running based system modeled after the IEC 61499 standard will be in parallel, Robust distributed system Incremental Cost suited for industry application and can be executable on the Estimation (RICE) algorithm, was introduced by authors to target platforms. The authors tested the proposed model on a handle the issue. The model contains a Gossip algorithm to 5-node system with industrial controllers. find the power mismatch estimation and consensus algorithm A combination of MAS and Particle Swarm Optimization for the incremental cost estimation. They report that their (PSO) called MAPSO (Multi-Agent Particle Swarm method outperforms the consensus method to packet loss Optimization) was proposed by C. Wu et al. [74] and was and delivered good results even with a 5% information loss applied to the ED problem. The proposed method overcomes in the network. Another study on distributed ED under com- the shortcomings of PSO, the fast convergence to the local munication uncertainties is by G. Wen et al. [80]. The pro- optima, and achieves high convergence speed and precision. posed approach utilized a robust consensus model to counter The agents are modeled to have the ability of self-learning to the communication uncertainties. A study of consensus-based improve the problem-solving ability. The authors verified the ED model under dynamic communication network is evaluat- effectiveness of the proposed method on IEEE test buses and ed by M. Hamdi et al. in [81]. T. Yang et al. [82]explored a the method was found to be faster than evolutionary algo- distributed ED model for a system with potentially time- rithms. A hybrid of MAS with PSO, deterministic search varying topologies and network delays. The authors proposed and bee decision-making process called HMAPSO (Hybrid a gradient push-sum based method to handle the network Multi-Agent based Particle Swarm Optimization) was pro- challenges. posed by R. Kumar et al. [75]. The HMAPSO method was The application of distributed ED using consensus the- applied to an ED model with valve-point effect and was ob- oryinamicrogridisbyproposed R. Wang et al.[83]. The served to be more robust and accurate than other PSO proposed method is a fully distributed approach without a methods. leader or a virtual control node. The incremental cost of 15 Page 8 of 15 Technol Econ Smart Grids Sustain Energy (2018) 3:15 each bus in the system is taken as the consensus variable. communication in the system. The decrease mobile agent The authors validated the performance and convergence (DA) are intended to reduce the generated output and of the distributed ED in a microgrid model. A similar increase mobile agents (UA) initiates an increase in gen- approach for distributed ED in an islanded microgrid is erated output. In the proposed approach, mobile agents proposed by Z. Tang et al. [84]byusing IC as thecon- travel throughout the system and negotiates with the gen- sensus variable. A distributed power dispatch model for a erator agents, depending up on the operating conditions. multi-microgrid scenario is reported by X. He et al. [85] The performance comparison of the method with dynamic utilizing a primal-dual consensus algorithm. They evalu- programming yielded similar results but it is not a fully ated the performance of the proposed algorithm on an decentralized method since it has a leader and mobile IEEE 30, 57 and 300 bus systems. A study on distributed nodes as an interface between generator agents. Figure 4 control architecture for a hybrid AC/DC microgrid is per- shows the architecture of the proposed method. formed by P. Lin et al. in [86]. In most of the above cited An improved version of the method proposed in [31] papers for ED, an evaluation of generation cost is not was presented by J. Yu et al. in [88]. The proposed MAS performed for the different approaches and moreover the agents are more intelligent and have the capability to problem of ED is not solved in a more realistic manner solve complex optimization problems. The profit maximi- with non-negligible losses. Limited work has been done zation objective is obtained using three types of agents, on the implementation cost of these approaches in a smart namely central agent, mobile agent and generator agent. grid system. A summary of the above cited papers and The central operator in the system is the central agent their features are listed in Table 3. which commands the mobile agents to achieve the objec- tive function. The mobile agents travel to each generator to negotiate and reach a satisfactory result. The model is Unit Commitment not fully decentralized as there is an agent acting as a central controller. The authors validated the proposed Unit Commitment is the process of determining the schedule model against a hybrid Lagrange Relaxation - Evolutio of generating units within a power system. The optimized nary Programming (LR - EP) method. schedule is generated subject to device and operating con- A MAS based UC model for a smart grid system consid- straints of the system with the objective of minimizing the cost ering RES uncertainty was developed by X. Zhang et al. [89]. for utilities [92]. The ED optimization is usually performed on The model considered the uncertainty associated with wind, the committed generators from the UC step. Various ap- solar and load along with the charging and discharging of proaches were used to find the optimal schedule from the PHEVs (plug-in hybrid electric vehicles). The hierarchical system consisted of management agents, work agents, coop- UC problem ranging from highly complex and theoretical methods to simple rule of thumb methods [93–97]. The scope erative co-evolution agent and object cooperative agent. The of the UC problem depends on the generation mix, operating work agents used adaptive GA to solve the optimization. and security constraints set by the utility. The focus of this A MAS based approach to solve the profit based UC review is on the decentralized approaches which utilized problem was explored by J. Yu et al. [99]. Rule-based, MAS to solve UC problem. and dynamic programming methods were used to solve Authors in [87] developed a centralized approach to solve the profit based UC. D. Sharma et al. [91] introduced an UC in a smart grid system using a MAS based architecture. improved version for the profit based UC. The ISO agents The agents communicate information to neighboring agents, in the proposed method used a rule-based intelligence to but the UC happens in a centralized controller. Figure 3 shows work in conjunction with generator agents to maximize the different agents in a centralized UC. The proposed method the objective function. The functionality of generator helps to reduce the communication overhead but has the de- agents is limited to maximize their profit for a given de- merits of a centralized controller such as a single point of mand and reserve using real – parametric Genetic failure, increased computation time with complexity, unavail- Algorithm and to share the information with the ISO ability of plug-and-play functionality etc. agents. The maximum profit generating agents are com- A distributed UC model based on MAS was developed mitted to the system by ISO agents while satisfying the by T. Nagata et al. [31, 98]. The proposed model consists up/down time constraints. The authors reported the per- of three types of agents namely Generator Agents (GA), formance of the proposed method with several hybrid mobile agents (DA, UA) and Facilitator Agent (FA). The methods. FA contains the objective function. The system level con- T. Logenthiran et al. [2] utilized MAS concept to de- straints are satisfied by the interaction between GA and velop a resource scheduling model for an islanded power mobile agents and the GA handles the local constraints. grid with integrated microgrids and DER. The proposed methodology has three stages; microgrid scheduling to The two mobile agents are provided to improve the Technol Econ Smart Grids Sustain Energy (2018) 3:15 Page 9 of 15 15 Table 3 High-level systematic summary of MAS application in ED and UC Problem/ Algorithm Platform Constraints Implementation Author, Year Remarks Architecture ED/Distributed Incremental Cost Consensus (ICC) – Demand Constraint Simulation on 3 unit and 5 unit Z. Zhang et al. + Successful implementation of system [57] consensus algorithm in ED 2011 - Leader node need to be selected Average consensus + ICC – Demand, Generator constraints Simulation on a 5 unit system Z. Zhang et al. + Not a leader-follower structure [59] + Computation more distributed Consensus based – Demand, Generator constraints 5 generators in a prototype N. Cai et al. [7] + utilized Local information among microgrid 2012 agents + proposed an improved communication algorithm for the agents Consensus based MATLAB Demand, Generator constraints Simulation on IEEE 14 bus S. Yang et al. + utilized mismatch between demand and system [60] power generation 2013 + same precision as the centralized method (Lambda-Iteration approach) Incremental Cost Consensus (ICC) MATLAB Demand Constraint 5 Node system (IEC 61499 G. Zhabelova + Developed an industrial model for the Architecture) et al. [73] MAS based ED 2013 + used industrial standard IEC 61499 Consensus based-parallel MATLAB Demand, generation, Transmission loss Simulation on IEEE 6 and 300 G. Benetti et al. +usedtwo consensus algorithms in constraints bus system [61] parallel 2014 + communication network has the same topology of the power system Dynamic average Consensus – Power and Generator constraints 15- Bus System A. Cherukuri + improved distributed coordination et al. [62] algorithm 2014 + can handle intermittent energy sources, dynamic load conditions Consensus based MATLAB/Simulink Generator, ramp-rate limit, line-flow Simulation of a five-generator J. Cao et al. [66] + considered generator dynamics limit constraints smart grid model 2014 + compared different communication topologies Distributed auction-based MATLAB Valve point effect, multiple fuel options Simulation on 10, 15 and 40 G. Binetti et al. + ability to solve non-convex ED algorithm and prohibited operating zones generators [71] problems 2014 +Utilizes a leaderless consensus protocol +More robust & Fault tolerant Distributed average consensus DICE /MATLAB Demand, Simulation on IEEE 118 and V. Loia et al. + self-organizing dynamic agents Generator limits 300 bus system [76] 2014 + A prototype version of the self-organizing architecture is developed Distributed replicator dynamics JADE Demand, Simulation on system with 3 K. Luo et al. + can be deployed for real-time Generator constraints DGs and 3 loads [63] applications 2015 + Fully distributed computation + utilized game-theory concepts Multi-Agent PSO MATLAB Demand, Simulation on IEEE 3, 13 and C. Wu et al. [74] + used a combination of MAS and Generator constraints, Valve-point effect 40 units 2015 Particle Swarm Optimization + faster than evolutionary algorithms Projected Gradient and Finite time – Demand, Generator constraints Simulation on 6 bus and IEEE F. Guo et al. + considered random wind power average consensus 30 bus system [65] generation 2016 + Incremental cost of generators not required 15 Page 10 of 15 Technol Econ Smart Grids Sustain Energy (2018) 3:15 Table 3 (continued) Problem/ Algorithm Platform Constraints Implementation Author, Year Remarks Architecture + Plug-and-Play property Consensus based – Demand, Generator constraints Simulation on a Five-unit Z. Yang et al. + virtual incremental cost as the system [67] consensus variable 2016 + maintains the supply-demand balance during transients. Consensus based- two protocols – Electrical and heat power balance Simulation on 16 Bus test Y. Li et al. [68] + modeled ED for a combined heat and constraints, capacity limits system 2016 power system + No leader or central controller Consensus based – Demand, Simulation on IEEE 57 Bus C. Li et al. [78] + event triggered consensus-based ED Generator constraints system 2016 model + reduces the communication burden in the network Robust distributed system – Demand, Simulation on IEEE 9 Bus Y. Zhang et al. + combination of two algorithms running incremental cost estimation Generator constraints system [79] in parallel (RICE) 2016 + can remain robust under information loss Consensus based- two in parallel – active and reactive power balance Simulation on IEEE 30 Bus P.P. Vergara + considered both active and reactive constraints system et al. [64] power in the model 2017 + based on primal-dual constrained decomposition Consensus based – Demand, Simulation on IEEE 14, 118 Z. Yang et al. + Does not rely on the supply-demand Generator constraints Bus systems [69] mismatch 2017 + Can be used in real-time applications Consensus based – Demand, Generator constraints Simulation on 10 bus and IEEE R. Wang et al. + Implemented distributed ED in a 118 bus system [83] microgrid 2018 + Considered bus power UC/ Centralized Centralized decision-making JADE Demand, renewable, storage constraints Energy transport network in S. Hajjar et al. + Energy storage considered algorithm France (RTE) [87] + reduced communication overhead UC/Hierarchical Simple negotiation strategies Java Demand, Reserve, Ramp-limit Simulation on 10 unit system T. Nagata et al. + obtained results close to dynamic constraints [31] programming dynamic programming and rule- Java Power and reserve limit, system Simulation on 10 unit system J. Yu et al. [88] + agents can solve complex optimization based method constraints 2005 problems Cooperative co-evolution algorithm Mathematica PHEV limit, 10-unit system with standard X. Zhang et al. + considered uncertainty associated with Generation constraints, Ramp rate limits input data of power plants [89] wind, solar and load 2016 + Considered PHEVs UC/ Distributed Zonal optimization approach MATLAB Demand constraints Realistic grid with two E. Kaegi et al. + Two-stage optimization model distributed energy sources [90] + There is no leader node in the model Real-parametric genetic algorithm JADE Demand, ramp limit, reserve constraints Simulation on a 10 unit system D. Sharma et al. + utilized genetic algorithm to solve the (GA) [91] optimization problem 2011 + Rule based intelligence added to ISO agents Technol Econ Smart Grids Sustain Energy (2018) 3:15 Page 11 of 15 15 Fig. 3 Smart grid communicating agents in a centralized UC [87] satisfy its internal demand as the first step, the second zone to reach the profit maximization. In the next stage, stage being contacting the network to analyze the possi- the optimization happens during the interaction between bility of exporting power, and the final step is to schedule zones. The paper only focused on the intra zonal activ- the whole microgrid considering both internal demand ities, but the inter-zonal activities also play a significant and the power transfer from the second stage. The authors role in profit maximization. Figure 5 represents the zon- al approach used by the authors. A summary of the used the JADE platform to simulate the MAS system and used Lagrangian Relaxation with Genetic Algorithm to work done by different researchers on ED and UC prob- schedule the microgrid resources internally. They report lem which utilized MAS concepts are summarized in the robustness and scalability of the method by testing it the Table 3. in a PoolCo energy market. E. Kaegietal. [90] proposed a decentralized ap- proach to solve the UC problem using the MAS con- Conclusion cept. The methodology was based on zonal approach consisting of generator agents, load agents and zone This paper presented a literature review on the applica- agents. The generator agents (GA) and load agents tion of MAS for ED and UC in a smart grid. The inte- (LA) handle the local profit maximization within a zone gration of DERs into a grid requires a decentralized con- while the zone agent handles the interaction with other trol strategy to incorporate these resources and to main- zones. The zone agents have no financial objectives but tain the grid resiliency. The multi-agent technology is a only acts as a service agent for the entities within its promising and scalable platform to implement distributed zone. The optimization is done in two stages. The intra resource scheduling and allocation using various compu- zone level is the competition between agents within a tational techniques. Fig. 4 MAS based UC model [31] 15 Page 12 of 15 Technol Econ Smart Grids Sustain Energy (2018) 3:15 Fig. 5 Agent hierarchy in the UC Model [90] Though there are many centralized algorithms being model which considers the increased DER penetration into the used to solve the ED problem, a small change in the smart future grid. Most of the reviewed articles focused on the im- grid may lead to redesign of these centralized approaches. plementation and convergence ability of the proposed Thus, there is a need for a distributed ED approach which methods, more work needs to be done on evaluating these can enjoy the benefit of robustness, scalability and less methods for their speed and cost savings in a real -time information requirement. Different distributed algorithms environment. for solving ED problem have been proposed by many We believe that this paper can act as a resource for re- researchers in the literature. Of all these distributed ap- searchers in academia and analysts in utilities to understand proaches, consensus-based algorithm has evolved as the the background on MAS’s application for smart grid manage- promising computing method for solving ED. The ment and control. consensus-based ED algorithm can make the analysis Acknowledgments The authors acknowledge the support of the National tractable by simplifying the system into linear for the it- Science Foundation (NSF) award #1537565 for this work. eration process. Most of the consensus-based algorithms available in the literature are useful in solving only con- Open Access This article is distributed under the terms of the Creative vex ED problem without transmission losses. On the other Commons Attribution 4.0 International License (http:// hand, an auction-based algorithm has been proposed to creativecommons.org/licenses/by/4.0/), which permits unrestricted use, solve nonconvex ED problem. However, most of the in- distribution, and reproduction in any medium, provided you give appro- vestigations reported in the literature are limited to imple- priate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. mentation in the simulation environment without address- ing the challenges of different scenarios of a smart grid in real time. Hence, these approaches have to be established References in real time which would be helpful in solving ED prob- lem in a smart grid. 1. Xu Y, Zhang W, Wenxin L (2015) Distributed Dynamic Researchers have explored the application of MAS in cen- Programming-Based Approach for Economic Dispatch in Smart tralized, hierarchical and distributed models for an UC prob- Grids. IEEE Trans Ind Informatics 11:166–175. https://doi.org/10. lem. Most recent works focused on a distributed model utiliz- 1109/TII.2014.2378691 ing concepts from the negotiation strategies and genetic algo- 2. Logenthiran T, Srinivasan D, Khambadkone AM (2011) Multi- agent system for energy resource scheduling of integrated rithm. The distributed UC models used a zonal approach microgrids in a distributed system. Electr Power Syst Res 81: consisting of controller/zonal agents facilitating the commu- 138–148. https://doi.org/10.1016/j.epsr.2010.07.019 nication among agents and these models are not completely 3. Arriagada E, Lopez E, Lopez M et al (2015) A probabilistic eco- distributed. There is a need for a completely distributed UC nomic dispatch model and methodology considering renewable Technol Econ Smart Grids Sustain Energy (2018) 3:15 Page 13 of 15 15 energy, demand and generator uncertainties. Electr Power Syst Res 22. Tolbert LM, Qi H, Peng FZ (2001) Scalable multi-agent system for 121:325–332. https://doi.org/10.1016/j.epsr.2014.11.018 real-time electric power management. 2001 Power Eng Soc 4. Nair AS, Ranganathan P, Kaabouch N (2017) A constrained topo- Summer Meet Conf Proc (Cat No01CH37262) 3:1–4. https://doi. org/10.1109/PESS.2001.970327 logical decomposition method for the next-generation smart grid. In: 2017 Int. Conf. Electr. Comput. Commun. Technol. (IEEE 23. Zheng G, Li N (2010) Multi-Agent Based Control System for ICECCT 2017). Coimbatore, pp 1–6. https://doi.org/10.1109/ Multi-Microgrids. 2010 Int Conf Comput Intell Softw Eng. ICECCT.2017.8118053 https://doi.org/10.1109/CISE.2010.5676818 5. Sujil A, Verma J, Kumar R (2016) Multi agent system: concepts, 24. Wang H (2001) Multi-agent co-ordination for the secondary voltage platforms and applications in power systems. Artif Intell Rev 1–30. control in power-system contingencies. IEE Proc - Gener Transm https://doi.org/10.1007/s10462-016-9520-8 Distrib 148:61–66. https://doi.org/10.1049/ip-gtd:20010025 6. Wooldridge M (2009) An introduction to multiagent systems. 25. Xu Y, Liu W, Gong J (2011) Stable multi-agent-based load shed- https://www.cs.ox.ac.uk/people/michael.wooldridge/pubs/imas/ ding algorithm for power systems. IEEE Trans Power Syst 26: IMAS2e.html. Accessed 5 Jan 2018 2006–2014. https://doi.org/10.1109/TPWRS.2011.2120631 7. Cai N, Nga NTT, Mitra J (2012) Economic dispatch in microgrids 26. Eddy YSF, Gooi HB, Chen SX (2015) Multi-agent system for dis- using multi-agent system. 2012 North Am Power Symp NAPS tributed management of microgrids. IEEE Trans Power Syst 30:24– 2012. https://doi.org/10.1109/NAPS.2012.6336435 34. https://doi.org/10.1109/TPWRS.2014.2322622 8. Bou Ghosn S, Ranganathan P, Salem S, et al (2010) Agent-oriented 27. Abedini R, Pinto T, Morais H, Vale Z (2013) Multi-agent approach designs for a self healing smart grid. 2010 First IEEE Int Conf for power system in a smart grid protection context. 2013 IEEE Smart Grid Commun. https://doi.org/10.1109/SMARTGRID. Grenoble Conf PowerTech, POWERTECH 2013. https://doi.org/ 2010.5622085 10.1109/PTC.2013.6652158 9. Leeton U, Kulworawanichpong T (2012) Multi-Agent Based 28. Wan H, Wong KP, Chung CY (2008) Multi-agent application in Optimal Power Flow Solution. In: 2012 Asia-Pacific Power protection coordination of power system with distributed genera- Energy Eng. Conf. pp 1–4. https://doi.org/10.1109/APPEEC. tions. IEEE Power Energy Soc 2008 Gen Meet Convers Deliv 2012.6307223 Electr Energy 21st Century, PES. https://doi.org/10.1109/PES. 10. Mohammadi J, Hug G, Kar S (2016) Agent-based distributed secu- 2008.4596261 rity constrained optimal power flow. IEEE Trans Smart Grid 9: 29. Do Nascimento LL, Rolim JG (2013) Multi-agent system for adap- 1118–1130. https://doi.org/10.1109/TSG.2016.2577684 tive protection in microgrids. 2013 IEEE PES Conf Innov Smart 11. Ren F, Zhang M, Soetanto D, Su X (2012) Conceptual design of a Grid Technol ISGT LA 2013. https://doi.org/10.1109/ISGT-LA. multi-agent system for interconnected power systems restoration. 2013.6554435 IEEE Trans Power Syst 27:732–740. https://doi.org/10.1109/ 30. Yong L, Tao S (2007) Economic dispatch of power system incor- TPWRS.2011.2177866 porating wind power plant. 2007 Int Power Eng Conf (IPEC 2007) 12. Nagata T, Sasaki H (2002) A multi-agent approach to power system 159–162. https://ieeexplore.ieee.org/document/4510019. Accessed restoration. IEEE Trans Power Syst 17:457–462. https://doi.org/10. 5 Jan 2018 1109/TPWRS.2002.1007918 31. Nagata T, Ohno M, Sasaki H, Fujita H (2002) A multi-agent ap- 13. Solanki JM, Khushalani S, Schulz NN (2007) A multi-agent solu- proach to unit commitment. Electr Eng Japan 141:41–47. https:// tion to distribution systems restoration. IEEE Trans Power Syst 22: doi.org/10.1002/eej.10057 1026–1034. https://doi.org/10.1109/TPWRS.2007.901280 32. Li N, Singhal NG, Hedman KW (2017) An enhanced security- 14. Khamphanchai W, Pisanupoj S, Ongsakul W, Pipattanasomporn M constrained unit commitment model with reserve response set pol- (2011) A multi-agent based power system restoration approach in icies. Proc. 50th Hawaii Int Conf Syst Sci. https://doi.org/10.24251/ distributed smart grid. Util Exhib Power Energy Syst Issues hicss.2017.371 Prospect Asia (ICUE), 2011 Int Conf. https://doi.org/10.1109/ ICUEPES.2011.6497754 33. Kantamneni A, Brown LE, Parker G, Weaver WW (2015) Survey 15. Solanki JM, Schulz NN, Gao W (2005) Reconfiguration for resto- of multi-agent systems for microgrid control. Eng Appl Artif Intell ration of power systems using a multi-agent system. Power Symp 45:192–203. https://doi.org/10.1016/j.engappai.2015.07.005 2005 Proc 37th Annu North Am. https://doi.org/10.1109/NAPS. 34. Dimeas AL, Hatziargyriou ND (2004) A multiagent system for 2005.1560571 microgrids. IEEE Power Eng Soc Gen Meet 2:55–58. https://doi. 16. Khamphanchai W, Pipattanasomporn M, Rahman S (2012) A org/10.1109/PES.2004.1372752 multi-agent system for restoration of an electric power distribution 35. Wang Z, Yang R, Wang L (2011) Intelligent multi-agent control for network with local generation. IEEE Power Energy Soc Gen Meet. integrated building and micro-grid systems. IEEE PES Innov Smart https://doi.org/10.1109/PESGM.2012.6345106 Grid Technol Conf Eur ISGT Eur. https://doi.org/10.1109/ISGT. 17. Raju L, Morais AA, Rathnakumar R et al (2017) Micro-grid Grid 2011.5759134 Outage Management using Multi Agent Systems. Energy Procedia 36. Lü P, Zhao J, Yao J, Yang S (2017) A decentralized approach for 117:112–119. https://doi.org/10.1016/j.egypro.2017.05.113 frequency control and economic dispatch in smart grids. IEEE J 18. Trigo P, Marques P (2008) The electricity market as a multi-agent Emerg Sel Top Circuits Syst 7:447–458. https://doi.org/10.1109/ system. 2008 5th Int Conf Eur Electr Mark EEM. https://doi.org/10. JETCAS.2017.2708900 1109/EEM.2008.4579092 37. Nygard KE, Ghosn SB, Mcculloch R, et al (2012) Decision support 19. Logenthiran T, Srinivasan D (2011) Multi-agent system for market independence in a smart grid. ENERGY 2012 Second Int Conf based microgrid operation in smart grid environment. Dept. Elec. Smart Grids, Green Commun IT Energy-aware Technol Decis Comp. Sci. National University of Singapore 38. Silvestre D, Hespanha JP, Silvestre C (2018) Broadcast and gossip 20. Kok K (2010) Multi-agent coordination in the electricity grid, from stochastic average consensus algorithms in directed topologies. concept towards market introduction. Proc 9th Int Conf Auton IEEE Trans Control Netw Syst. https://doi.org/10.1109/TCNS. Agents Multiagent Syst Ind track 2018.2839341 21. Santos G, Pinto T, Morais H et al (2015) Multi-agent simulation of 39. Raffaelli G, Marsili M (2005) Statistical mechanics model for the competitive electricity markets: Autonomous systems cooperation emergence of consensus. Phys Rev E Stat Nonlinear Soft Matter for European market modeling. Energy Convers Manag 99:387– Phys 72. https://doi.org/10.1103/PhysRevE.72.016114 399. https://doi.org/10.1016/j.enconman.2015.04.042 15 Page 14 of 15 Technol Econ Smart Grids Sustain Energy (2018) 3:15 40. Gu D, Wang Z (2009) Leader-follower flocking: Algorithms and 60. Yang S, Tan S, Xu J-X (2013) Consensus Based Approach for experiments. IEEE Trans Control Syst Technol 17:1211–1219. Economic Dispatch Problem in a Smart Grid. IEEE Trans Power https://doi.org/10.1109/TCST.2008.2009461 Syst 28:4416–4426. https://doi.org/10.1109/TPWRS.2013. 41. Briñón-Arranz L, Schenato L (2013) Consensus-based source- seeking with a circular formation of agents. In: Eur. Control Conf. 61. Binetti G, Naso D, Turchiano B, et al (2014) Consensus-based pp 2831–2836. https://doi.org/10.23919/ECC.2013.6669782 approach for the economic dispatch problem. In: Proc. IFAC 42. Zhang X, Xu H, Yu T et al (2016) Robust collaborative consensus World Congr. IFAC, pp 3140–3145. https://doi.org/10.3182/ algorithm for decentralized economic dispatch with a practical com- 20140824-6-ZA-1003.01786 munication network. Electr Power Syst Res 140:597–610. https:// 62. Cherukuri A, Cortés J (2014) Distributed coordination for econom- doi.org/10.1016/j.epsr.2016.05.014 ic dispatch under varying loads and generator commitment. Fifty- 43. Hug G, Kar S, Wu C (2015) Consensus+Innovations Approach for second Annu Allert Conf. https://doi.org/10.1109/allerton.2014. Distributed Multiagent Coordination in a Microgrid. IEEE Trans Smart Grid 6:1893–1903. https://doi.org/10.1109/TSG.2015. 63. Luo K, Wu Q, Jacob Ø, Hejde NA (2015) Real-time distributed economic dispatch for distributed generation based on multi-agent 44. Lü P, Zhao J, Yao J, Yang S (2017) A decentralized approach for system. 2015 Mod. Electr. Power Syst. https://doi.org/10.1109/ frequency control and economic dispatch in smart grids. IEEE J MEPS.2015.7477174 Emerg Sel Top Circuits Syst PP:1–12 64. Vergara PP, Shaker HR, Norregaard B, da Silva LCP (2017) 45. Pipattanasomporn M, Feroze H, Rahman S (2009) Multi-agent sys- Distributed consensus-based economic dispatch considering grid tems in a distributed smart grid: Design and implementation. operation. 2017 IEEE Power Energy Soc Gen Meet. https://doi. IEEE/PES Power Syst Conf Expo 2009 PSCE ‘09 1–8. https:// org/10.1109/PESGM.2017.8273864 doi.org/10.1109/PSCE.2009.4840087 65. Guo F, Wen C, Mao J, Song YD (2016) Distributed Economic 46. Kravari K, Bassiliades N (2015) A survey of agent platforms. Dispatch for Smart Grids with Random Wind Power. IEEE Trans JASSS. https://doi.org/10.18564/jasss.2661 Smart Grid 7:1572–1583. https://doi.org/10.1109/TSG.2015.2434831 47. Singh A, Juneja D, Sharma AK (2011) Agent Development 66. Cao J, Yu M, Tung LJ (2014) Consensus-based distributed control Toolkits. Int J Adv Technol 2:158–164 for economic dispatch problem with comprehensive constraints in a 48. Shoham Y (1991) AGENT0: A simple agent language and its in- smart grid. EAI Endorsed Trans Energy Web 1:1–9. https://doi.org/ terpreter. In: Proc. AAAI-91. pp 704–709 10.4108/ew.1.2.e3 49. Thomas SR (1994) The PLACA agent programming language. In: 67. Yang Z, Xiang J, Li Y (2016) Distributed virtual incremental cost Intell. Agents. pp 355–370 consensus algorithm for economic dispatch in a microgrid. IEEE Int 50. Bellifemine F, Caire G, Greenwood D (2007) Developing multi- Conf Control Autom ICCA. https://doi.org/10.1109/ICCA.2016. agent systems with JADE. John Wiley Sons Ltd. https://doi.org/10. 1002/9780470058411 68. Li Y-S, Zhang H-G, Huang B-N, Teng F (2016) Distributed optimal 51. Gutknecht O, Ferber J (2001) The MadKit agent platform architec- economic dispatch based on multi-agent system framework in com- ture. In: Infrastruct. Agents, Multi-Agent Syst. Scalable Multi- bined heat and power systems. Appl Sci. https://doi.org/10.3390/ Agent Syst. SE - 5. pp 48–55 app6100308 52. Guo Q, Han J, Yoon M, Jang G (2012) A study of economic dis- 69. Yang Z, Xiang J, Li Y (2017) Distributed consensus based supply- patch with emission constraint in smart grid including wind turbines demand balance algorithm for economic dispatch problem in a and electric vehicles. 2012 IEEE Veh Power Propuls Conf VPPC smart grid with switching graph. IEEE Trans Ind Electron 64: 2012 1002–1005. https://doi.org/10.1109/VPPC.2012.6422753 1600–1610. https://doi.org/10.1109/TIE.2016.2615037 53. Nair AS, Ranganathan P, Salehfar H, Kaabouch N (2017) 70. Xing H, Mou Y, Fu M, Lin Z (2014) Consensus based bisection Uncertainty quantification of wind penetration and integration into approach for economic power dispatch. Proc IEEE Conf Decis smart grid: a survey. North Am. Power Symp. https://doi.org/10. Control 2015–Febru. https://doi.org/10.1109/CDC.2014.7039979 1109/NAPS.2017.8107196 71. Binetti G, Davoudi A, Naso D, et al (2014) A distributed auction- 54. Dominguez-Garcia AD, Cady ST, Hadjicostis CN (2012) based algorithm for the nonconvex economic dispatch problem. In: Decentralized optimal dispatch of distributed energy resources. IEEE Trans. Ind. Informatics 10:1124–1132. https://doi.org/10. In: Proc. IEEE Conf. Decis. Control. pp 3688–3693. https://doi. 1109/TII.2013.2287807 org/10.1109/CDC.2012.6426665 72. Binetti G, Davoudi A, Lewis FL et al (2014) Distributed consensus- 55. Rana R, Bhattacharjee S, Mishra S A consensus based solution to based economic dispatch with transmission losses. IEEE Trans the classic economic dispatch problem on a multi-agent system Power Syst 29:1711–1720. https://doi.org/10.1109/TPWRS.2014. framework. In: 2018 IEEMA Eng. Infin. Conf. IEEE, pp 1–6. https://doi.org/10.1109/ETECHNXT.2018.8385372 73. Zhabelova G, Vyatkin V, Zhang Z, Chow MY (2013) Agent-based 56. Yu W, Li C, Yu X et al (2018) Economic power dispatch in smart distributed consensus algorithm for decentralized economic dis- grids: a framework for distributed optimization and consensus dy- patch in Smart Grid. In: IECON Proc. (Industrial Electron. Conf. namics. Sci China Inf Sci 61:1–16. https://doi.org/10.1007/s11432- Vienna, Austria. https://doi.org/10.1109/IECON.2013.6699433 016-9114-y 74. Wu C, Li H, Wu L, Wu Z (2015) A Multi-Agent Particle Swarm 57. Zhang Z, Chow MY (2011) Incremental cost consensus algorithm Optimization for Power System Economic Load Dispatch. J in a smart grid environment. IEEE Power Energy Soc Gen Meet. Comput Commun 3:83–89 https://doi.org/10.1109/PES.2011.6039422 75. Kumar R, Sharma D, Sadu A (2011) A hybrid multi-agent based 58. Zhang Z, Chow M (2012) Convergence Analysis of the Incremental particle swarm optimization algorithm for economic power dis- Cost Consensus Algorithm Under Different Communication patch. Int J Electr Power Energy Syst 33:115–123. https://doi.org/ Network Topologies in a Smart Grid. IEEE Trans Power Syst 27: 10.1016/j.ijepes.2010.06.021 1761–1768 59. Zhang Z, Ying X, Chow MY (2011) Decentralizing the economic 76. Loia V, Vaccaro A (2014) Decentralized economic dispatch in smart dispatch problem using a two-level incremental cost consensus al- grids by self-organizing dynamic agents. IEEE Trans Syst Man, gorithm in a smart grid environment. NAPS 2011 - 43rd North Am Cybern Syst 44:397–408. https://doi.org/10.1109/TSMC.2013. Power Symp. https://doi.org/10.1109/NAPS.2011.6025103 2258909 Technol Econ Smart Grids Sustain Energy (2018) 3:15 Page 15 of 15 15 77. Zhao C, He J, Cheng P, Chen J (2017) Analysis of Consensus- 88. Yu J, Zhou J, Hua B, Liao R (2005) Optimal Short-Term Generation Based Distributed Economic Dispatch Under Stealthy Attacks. Scheduling with Multi-Agent System under a Deregulated Power. IEEE Trans Ind Electron 64:5107–5117 Int J Comput Cogn 3:61–65 89. Zhang X, Xie J, Zhu Z et al (2016) Smart grid cost-emission unit 78. Li C, Yu X, Yu W et al (2016) Distributed event-triggered scheme commitment via Co-evolutionary agents. Energies 9:1–13. https:// for economic dispatch in smart grids. IEEE Trans Ind Informatics doi.org/10.3390/en9100834 12:1775–1785. https://doi.org/10.1109/TII.2015.2479558 90. Kaegi E, Cherkaoui R, Germond A (2008) Decentralized unit com- 79. Zhang Y, Rahbari-Asr N, Chow MY (2016) A robust distributed mitment and dispatch for the distribution systems using intelligent system incremental cost estimation algorithm for smart grid econom- agents approach. In: PSCC 2008 ic dispatch with communications information losses. J Netw Comput 91. Sharma D, Srinivasan D, Trivedi A (2011) Multi-agent approach for Appl 59:315–324. https://doi.org/10.1016/j.jnca.2015.05.014 profit based unit commitment. In: 2011 IEEE Congr. Evol. Comput. 80. Wen G, Yu X, Liu ZW, Yu W (2017) Adaptive Consensus-Based CEC 2011. https://doi.org/10.1109/CEC.2011.5949932 Robust Strategy for Economic Dispatch of Smart Grids Subject to 92. Padhy NP (2004) Unit Commitment—A Bibliographical Survey. Communication Uncertainties. IEEE Trans Ind Informatics 14: IEEE Trans Power Syst 19:1196–1205. https://doi.org/10.1109/ 2484–2496. https://doi.org/10.1109/TII.2017.2772088 TPWRS.2003.821611 81. Hamdi M, Chaoui M, Idoumghar L, Kachouri A (2018) 93. Carrión M, Arroyo JM (2006) A computationally efficient mixed- Coordinated consensus for smart grid economic environmental integer linear formulation for the thermal unit commitment prob- power dispatch with dynamic communication network. IET lem. IEEE Trans Power Syst 21:1371–1378. https://doi.org/10. Gener Transm Distrib 12:2603–2613. https://doi.org/10.1049/iet- 1109/TPWRS.2006.876672 gtd.2017.1197 94. Lopez CJ, Ano O, Ojeda Esteybar D (2018) Stochastic Unit 82. Yang T, Lu J, Wu D et al (2017) A Distributed Algorithm for Commitment & Optimal Allocation of Reserves: A Hybrid Economic Dispatch over Time-Varying Directed Networks with Decomposition Approach. IEEE Trans Power Syst 8950:1–11. Delays. IEEE Trans Ind Electron 64:5095–5106. https://doi.org/ https://doi.org/10.1109/TPWRS.2018.2817639 10.1109/TIE.2016.2617832 95. Wang Y, Shahidehpour M, Lai LL et al (2018) Resilience- 83. Wang R, Li Q, Zhang B, Wang L (2018) Distributed Consensus Based Constrained Hourly Unit Commitment in Electricity Grids. IEEE Algorithm for Economic Dispatch in a Microgrid. IEEE Trans Smart Trans Power Syst 8950:1–10. https://doi.org/10.1109/TPWRS. Grid 0:1–11. https://doi.org/10.1109/TSG.2018.2833108 2018.2817929 84. Tang Z, Hill DJ, Liu T (2018) A Novel Consensus-Based Economic 96. Xiong P, Singh C (2018) A Distributional Interpretation of Dispatch for Microgrids. IEEE Trans Smart Grid 9:3920–3922. Uncertainty Sets in Unit Commitment under Uncertain Wind https://doi.org/10.1109/TSG.2018.2835657 Power. IEEE Trans Sustain Energy 3029:1–9. https://doi.org/10. 85. He X, Yu J, Huang T, Li C (2018) Distributed power management 1109/TSTE.2018.2828421 for dynamic economic dispatch in the multimicrogrids environ- 97. Abedi S, He M, Member S, et al (2018) Congestion risk-aware unit ment. IEEE Trans Control Syst Technol 1–8. https://doi.org/10. commitment with significant wind power generation. IEEE Trans 1109/TCST.2018.2816902 Power Syst 33:6861–6869. https://doi.org/10.1109/TPWRS.2018. 86. Lin P, Jin C, Xiao J et al (2018) A Distributed Control Architecture for Global System Economic Operation in Autonomous Hybrid 98. Nagata T, Ohono M, Kubokawa J et al (2002) A multi-agent approach AC/DC Microgrids. IEEE Trans Smart Grid 3053:1–13. https:// to unit commitment problems. Proc IEEE Power Eng Soc Transm doi.org/10.1109/TSG.2018.2805839 Distrib Conf 1:64–69. https://doi.org/10.1109/PESW.2002.984955 87. Hajjar S, Bratcu AI, Hably A (2015) A day-ahead centralized unit 99. Yu J, Zhou J, Wu W, et al (2004) Solution of the profit-based unit commitment algorithm for a multi-agent smart grid. In: Fed Conf commitment problem by using multi-agent system. In: WCICA Comput Sci Inf Syst. pp 265–271. https://doi.org/10.15439/ 2004. Fifth World Congr. Intell. Control Autom. https://doi.org/ 2015f227 10.1109/WCICA.2004.1343686

Journal

Technology and Economics of Smart Grids and Sustainable EnergySpringer Journals

Published: Oct 29, 2018

References