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

Learn More →

Design of the Exercise Load Data Monitoring System for Exercise Training Based on the Neural Network

Design of the Exercise Load Data Monitoring System for Exercise Training Based on the Neural Network Hindawi Journal of Healthcare Engineering Volume 2021, Article ID 7340140, 6 pages https://doi.org/10.1155/2021/7340140 Research Article Design of the Exercise Load Data Monitoring System for Exercise Training Based on the Neural Network Panlong Qin and Wei Feng Department of Physical Education, Hebei Academy of Fine Arts, Hebei,050700, China Correspondence should be addressed to Wei Feng; 20122474@stu.nun.edu.cn Received 7 August 2021; Accepted 16 September 2021; Published 25 September 2021 Academic Editor: Balakrishnan Nagaraj Copyright © 2021 Panlong Qin and Wei Feng. ,is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. In order to monitor the sports load data of athletes in sports training, this paper studies the methods and systems of sports load monitoring and fatigue warning based on neural network technology. In this paper, the neural network parallel optimization algorithm based on big data is used to accurately estimate the motion load and intensity according to the determined motion mode and acceleration data, so as to realize the real-time monitoring of the exercise training. ,e results show that the value of η is usually small to ensure that the weight correction can truly follow the direction of the gradient descent. In this paper, 176 samples were extracted from the monitoring data collected by the “National Tennis Team Information Platform,” 160 of which were selected as training samples and the other 16 as test samples. Ant colony size M � 20. ,e minimum value W of the weight min interval is −2, and the maximum value W is 2. ,e maximum number of iterations is set to 200. σ �1; that is, only one optimal max solution is retained. ,e domain is divided into 60 parts evenly; that is, r � 60. Generally, η can be taken as any number [28] between [10-3, 10], but the value is usually small to ensure that the weight correction can truly follow the direction of the gradient descent. In this paper, the value is 0.003. In the early warning stage of exercise fatigue, reasonable measurement units of exercise fatigue time were divided according to the characteristics of different exercise items. It is proved that the Bayesian classification algorithm can effectively avoid the sports injury caused by overtraining by warning the fatigue and preventing the sports injury caused by overtraining. (ACO) is a novel bionic evolutionary algorithm, which 1. Introduction shows excellent performance and great development po- Artificial neural network (ANN) is a kind of nonlinear tential in solving complex optimization problems, espe- dynamic system, which is borrowed from the development cially discrete optimization problems. ACO, as a global of biological neural network [1], the new intelligent in- optimization heuristic algorithm, is used to train the weight formation processing system. With its unique information of neural network, which can avoid the defect of BP neural processing methods, ANN has been widely used in various network. Nowadays, the competition of sports science and fields, such as industrial production monitoring, classifi- technology is increasingly fierce [3]. It is necessary to have cation, prediction and forecasting, analysis and processing, scientific training methods and means in order to improve auxiliary diagnosis of diseases, environmental pollution, sports performance. Sports competition is a big competi- and purification prediction. BP neural network is one of the tion of science and technology. ,e guiding ideology of representative neural network models and has been widely implementing competitive sports science and technology used. However, BP algorithm has some defects, such as low work is to carry out the thought that science and technology efficiency, slow convergence speed, and being easy to fall is the first productive force, carry out the strategy of “in- into local minimum, which restrict the application of BP vigorating sports through science and education,” and network in various fields [2]. Ant Colony Optimization perfect the management system and movement mechanism 2 Journal of Healthcare Engineering monitoring terminal in power transmission, adopts the that sports training and sports science and technology are closely combined [4]. Our research focuses on the actual measures of establishing identification table in the system and judging whether the data frame sent by the terminal needs in training, aiming at the key problems in training practice, to carry out scientific research and public rela- address has been sent before identification, so as to improve tions, and strengthen technical innovation. To promote the the real-time performance of the identification system [14]. scientific sports training, give full play to the leading role of Zhou et al. designed a neural network compensator based science and technology and improve the sports skills of our on the nonlinear error of the motion control system of the athletes and the competitive strength in the world com- image measuring instrument. ,e neural network simu- petition, for our athletes in the 2008 Olympic Games to get lator and compensator of nonlinear servo motor are trained good results to win the gold medal to make contributions by the data of input and output of servo motor. ,e [5]. designed neural network compensator is applied to the high precision positioning system of motion control, which makes the control system show good control performance. 2. Literature Review Simulation results show that the controller is effective [15]. In this paper, the detection data of physiological and bio- With the continuous maturity and development of artificial neural network technology, the intelligent characteristics chemical indexes of national athletes were taken as training samples, and the ACO-BP algorithm was adopted to inte- and capabilities of neural network are increasingly ex- grate ACO and BP algorithm to complete the neural network panded in its application fields. Many problems that cannot training [6, 7]. ,e ant colony neural network prediction be solved by traditional information processing methods model was established for sports load, and the corre- have achieved good results after using neural network, and sponding relationship between physiological and bio- especially in the engineering field it has been widely used. chemical indexes in sports and training load was obtained. With the continuous development of neural network theory itself and related theories and technologies, the Firstly, ACO is used to optimize the neural network weights as a whole to overcome the shortage of BP algorithm which application of artificial neural network will be more in depth and extensive. is easy to fall into local optimum [8, 9]. ,en taking the better weight as the initial value, BP algorithm is used to do In this paper, the test data of Chinese athletes’ physio- further optimization, to overcome the shortcomings of a logical and biochemical indexes are taken as training single ACO training network with long time and low pre- samples, and the ACO and BP algorithm are combined to cision [10–12]. complete the neural network training method, namely ACO- ,e research continues; Jeon and Kim used artificial BP algorithm, to establish the ant colony neural network intelligence and expert system design principle, and the prediction model for the exercise load and obtain the cor- method of simulated medical expert diagnosis, treatment of responding relationship between the physiological and disease thinking process, and compiled computer program biochemical indexes and the training load during the ex- ercise. Firstly, ACO is used to optimize the weight of the can help doctors to solve complex medical problem, as the doctor concluded that it is an important auxiliary tool for neural network globally to overcome the shortcoming of BP algorithm which is easy to fall into local optimal. ,en, with the treatment of disease and prognosis. Medical expert system is an important application of artificial intelligence the optimal weight as the initial value, BP algorithm is used in medicine. It is a collection of knowledge, computer to do further optimization, so as to overcome the disad- technology, network technology, communication tech- vantages of a single ACO training network such as long time nology, database technology, and medical science. ,e and low accuracy. development of medical expert system has become an important topic in modern medicine [13]. Cavina et al. 3. Research Methods designed a multiprotocol recognition system based on BP artificial neural network. ,e specific research contents of ,eoretical analysis proves that a feedforward net with a this paper include the design of multiprotocol recognition single hidden layer can map all continuous functions only eigenvalue, the data of a certain length of frame head is when learning. Two hidden layers are required when the taken as the eigencode, and the eigenvalue is obtained after function is discontinuous. Increasing the number of hidden normalized processing. ,e selection of multiprotocol layers can improve the nonlinear mapping ability of BP learning algorithm, simulation analysis, and comparison of network, but when the number of hidden layers exceeds a several BP neural network algorithm is according to the certain value, the performance of BP network will decline. In convergence rate and recognition rate of conjugate gradient 1998, Hecht-Nielson proved that a continuous function in descent method as the multiprotocol recognition system BP any closed interval can be approximated by a BP network neural network identification algorithm. As per the es- with a hidden layer. ,erefore, this paper will adopt a three- tablishment of multiprotocol recognition network, through layer BP network with a hidden layer, namely, R-N-1. Neural simulation, according to the recognition rate, the maxi- network structure model is where n is the number of input mum number of hidden layer of BP neural network suitable nodes, r is the number of hidden nodes, and the number of for the multiprotocol recognition system in this paper is 9. output nodes is 1. ,e number of neurons in the input layer ,e design of multiprotocol identification system, aiming of the neural network is determined by the influencing at the characteristic of the relatively fixed address of the factors [16, 17]. In this paper, the physiological and Journal of Healthcare Engineering 3 biochemical indexes which can obviously reflect the phys- 1 l iological function and the professional training years are taken as the influencing factors. ,e number of nodes in the output layer is set as 1, and the value range is [−1, 1], which is 2 1 used as the quantified value to measure the load. ,e closer the value is to 1, the smaller the load is, and vice versa. ,e role of hidden layer nodes is to extract and store the inherent rules from the samples. Each hidden layer node has several weights, and each weight is a parameter to enhance the mapping ability of the network. If the number of hidden Figure 1: BP network structure. nodes is too small, the network’s ability to obtain infor- mation from the sample is poor, which is not enough to normalized input and target data are subject to normal summarize and reflect the sample rule of the training distribution; i.e., sample. If the number of hidden nodes is too large, it may also learn to remember the irregular content of the sample, [pn, meanp, stdp] � prestd(p), (2) such as noise, which reduces the generalization ability. where pn is the input direction after vectorization, p is a Generally, the following empirical formula can be used group of collected data (vector), and meanp is the input for the number of hidden layer nodes in three-layer forward direction after vectorization. Enter the mean value of the network: vector, and STDP is the deviation of the input vector. ,e √���� � algorithm executed is m � n + 1 + a, (p − meanp) m � log 2 , (1) pn � . (3) √�� stdp m � nl. As physiological and biochemical indexes, some indexes M is the number of nodes in the hidden layer, N is the are correlated. ,rough the principal component analysis of number of nodes in the input layer,L is the number of nodes standardized data, the redundant components of sample in the output layer, and α is a constant between 1 and 10. In data can be eliminated; that is, the components that con- this paper, the number of hidden layer nodes is determined tribute little to the overall change of the data set, so as to according to equation (1). To sum up, the BP neural network reduce the data dimension. ,e processing function of adopted in this paper is shown in Figure 1. MATLAB is [Ptrans, TransMat] � prepca(pn, 0.02); (4) 3.1. Processing of Sample Data. Tennis sport has the longer where 0.02, the second parameter in the prepca function, game time, the intensity of the intermittent type, short smell, indicates that the main element is an element whose square and explosive force action composition. Tennis is more and deviation from the mean is greater than 2% of the variance of more towards the strength and speed against the direction of the vector. TransMat is a change matrix with major elements. development; the physical requirements are very high. With In this paper, the blood data and urine data were ana- the increase of professional training years, various functions lyzed by principal component analysis. ,rough the analysis of the body will change, and the ability to adapt to the load and processing of a group of processed blood test data and will also change. Athletes with short training years can bear urine data of national tennis team players, the urine sample less amount of exercise than those with long training years. data were changed from the original 9. ,e dimension (8 ,e same amount of exercise may be a large exercise load for dimensions of urine index and professional training years) young players. ,erefore, the professional training years can was reduced to 6, and the data of blood data were reduced to be used as an important influencing factor of load fore- 13 from 15 (14 dimensions of blood index and professional casting to improve the accuracy of assessment ability. training years). ,e training samples used in the experiment are from the tennis team sports collected by the “National Tennis Team Information Platform” physiological and biochemical in- 3.2. Basic Steps. In the case that the structure of the neural dexes monitoring information of the mobilized competition network has been determined, the basic steps of training the and daily training. Due to the size of the original test data, in neural network by the ACO-BP method are as follows: order to avoid the difference of physical meaning and unit of Step 1: initialize: input vector on the network model, the difference of bit is obvious. Before the training of neural network, the training ,e weights of all the weights of the interval [W and min samples need to be processed in advance, that is, normalized W ] are evenly divided intor subregions. Each region max processing [18, 19]. In order to train the function better, we that the points on the boundary is the value of an also can avoid the network failure caused by being too large. alternative for each parameter setting up a list of ,e numbers are collected in this article. ,e data are pheromones as shown in Table 1. At the initial moment, mapped between [−1, 1] and normalized, so that the set each point to have the same amount of pheromone 4 Journal of Healthcare Engineering Table 1: Pheromone table of weights. Step 7: the neural network is further trained by BP algorithm, the best weights of σ group found by ANT Label 1 2 ... r + 1 Divide the scale a1 a2 ... ar + 1 colony algorithm are taken as the initial weights of BP ,e pheromone values (1) (2) ... (r + 1) algorithm, the error between network output and actual output is calculated, and the error is propagated one- way from the output layer to the input layer to adjust tau 0, pheromone volatilization coefficients rho, and the weights. Repeat the process until the stop condi- time t, and the number of loop NC is zero. Set the tions are met. maximum number of loops of the ending condition, Step 8: use validation samples to test the generalization MAXNC. ability of the trained neural network. If the validation Step 2: release all M ants. Ant K uses the following error meets the requirements, quit the program. probability formula as a path selection rule. ,at is, Otherwise, go to Step 1 and start training again. select an element in each set Ipi(1≤I≤m). 4. Result Analysis τ(i) p (i) � . (5) In order to verify the network model, the author has done a 􏽐 τ(i) 1≤j≤m lot of computer simulation experiments. In women’s team sports, as an example, the blood index data of the team To record the label of the point passed by the ant, a members were collected from the monitoring data collected value is selected for the weight and recorded in K. After by the “National Tennis Team Information Platform.” 176 the ant selects the values for all the weight parameters, samples were taken, and 160 of them were selected as the ant completes a traversal, and all the values it training samples and the other 16 were selected as test chooses to record constitute all the parameters of the samples. Ant colony size M � 20. ,e minimum value of the neural network. weight interval W is −2, and the maximum value W is min max Step 3: repeat Step 2 until all the ants reach the food 2. ,e maximum number of iterations is set to 200. σ �1; that source. is, only one optimal solution is retained. ,e domain is Step 4: let T← +M, NC← NC + 1. Use the weight evenly divided into 60 parts; that is, r � 60; η can generally parameters selected by each ant to calculate the take any number between [10-3, 10], but it usually takes a output results of the neural network and the error E, small value to ensure that the weight correction can really record the weight of σ group with small error and the descend along the direction of the gradient. In this paper, the current optimal solution Emin, and compare the size value is 0.003. ,e specific parameter settings of the network of Emin and E0. If Emin≤E0, go to Step 8; otherwise, are shown in Table 2. go to Step 5. According to the actual situation of the national tennis Step 5: update the pheromone of each element team and the need of preparing for the Olympic Games, the according to the pheromone regulation rules below. national tennis team information platform is designed and Pheromone regulation: as time goes by, the phero- developed with Delphi7.0 development tool and SQL mone that was left behind gradually fades away. ,e Server2000 database. ,e information platform includes volatility coefficient ρ represents the persistence of the athletes’ basic information, athletes’ training and competi- pheromone, and 1−ρ represents the vanishing degree tion information data, image and video analysis and training of the pheromone. After n time units, the ants travel monitoring data, which is a set of analysis and management from the nest to the food source, and the pheromones system for tennis. It mainly stores the basic information of along each path are adjusted according to the fol- Chinese tennis players, athletes’ injuries and rehabilitation, lowing formula: nutritional recovery information, athletes’ special energy, physiology and biochemistry, training load, training τ 􏼐I 􏼑(t + m) � pτ I 􏼁 (t) + Δτ 􏼐I 􏼑, j pi j Pi j pi method, training course content, training effect, competition process, competition results, and other training match in- ⎪ formation data. From the perspective of function, it mainly (6) k e includes data management, video image management, and Δτ I � 􏼐 􏼑 j pi data analysis. 0, ,e output results corresponding to the samples were jointly determined by experts, team doctors, and other where Q is a constant, used to adjust the pheromone people in combination with the corresponding training adjustment speed. e is the output error when a set of plans and athletes’ performance at that time. We use minus 1 weights selected by the KTH ant is taken as the weights for body function. Poor athletes cannot adapt to the current of the neural network and is defined as e � |o − o |, training plan and should immediately adjust the amount of where o ando are the actual and expected outputs of training. Let us denote the athlete by 0. Being able to adapt to the neural network. As you can see, the smaller the the current training plan, exercise amount is appropriate and error, the larger the corresponding pheromone can maintain the training plan for training; 1 indicates that increase. the body function is in a completely normal state, the Journal of Healthcare Engineering 5 Table 2: ACO-BP and BPNN parameter settings. ,e name of the M W W Q r NACO NBP EO min max ACO-BP 20 −2 2 0.005 60 200 0.003 12000 0.005 BPNN −0.1 0.1 0.003 20000 0.005 Table 3: Comparison of simulation test results. ,e mean square error of the Nth iteration ,e experimental method N � 100 N � 500 N � 1000 Standard BP algorithm 0.00977462 0.0063423 0.00429856 BP algorithm for adding the momentum term 0.00868283 0.00416346 0.000683606 ACO-BP algorithm 0.0071246 0.00198015 8.9427e − 5 athletes can fully adapt to the current amount of exercise, by warning the fatigue and preventing the sports injury and can appropriately increase the amount of training to caused by overtraining. ,e research of this paper is a improve the training effect. ,e test results obtained through preliminary study, combined with ability and the limitation the simulation test are listed in Table 3. of technical conditions. ,ere are a lot of improvement, From the above experimental results, the following mainly having the following several aspects: (1) the ant conclusions can be drawn: the three-layer front is based on colony algorithm is a new bionic algorithm. Now still not ant colony optimization algorithm presented in this paper. mature, the parameters of the ant colony algorithm selection Based on the training model of feedback neural network (BP theoretical guidance remain to be further strengthened, neural network), a new network training algorithm is because it directly relates to the parameter selection of established. ,e experimental results are analyzed and application effect of the algorithm. (2) ,e ant colony al- compared with the standard BP algorithm and the improved gorithm is further improved to increase the convergence of BP algorithm with momentum term, ACO-BP. ,e algo- the improved ant colony algorithm to meet the needs of rithm has a fast convergence speed and can achieve a smaller large-scale network. (3) Integrate qualitative and quantita- value of mean square error. ,erefore, the convergence tive problems, and quantify fuzzy concepts, so as to better process of this method has obvious advantages and stability. meet the needs of users. 5. Summary of This Paper Data Availability In order to monitor the sports load data of athletes in sports ,e data used to support the findings of this study are training, this paper studies the methods and systems of available from the corresponding author upon request. sports load monitoring and fatigue warning based on neural network technology. In this paper, the neural network Conflicts of Interest parallel optimization algorithm based on big data is used to accurately estimate the motion load and intensity according ,e authors declare that they have no conflicts of interest. to the determined motion mode and acceleration data, so as to realize the real-time monitoring of the exercise training. References ,e results show that the value of η is usually small to ensure that the weight correction can truly follow the direction of [1] M. A. Yan, H. Deng, and X. J. University, “Design of sports gradient descent. In this paper, 176 samples were extracted training correction system based on virtual reality,” Infor- from the monitoring data collected by the “National Tennis mation & Technology, vol. 28, 2019. Team Information Platform”, 160 of which were selected as [2] Y. Lin, Q. Qu, Y. Lin et al., “Customizing robot-assisted training samples and the other 16 as test samples. Ant colony passive neurorehabilitation exercise based on teaching sizeM � 20. ,e minimum value W of the weight interval min training mechanism,” BioMed Research International, is −2, and the maximum value W is 2. ,e maximum vol. 2021, no. 31, 10 pages, Article ID 9972560, 2021. max [3] A. A. Ibrahim, O. W. Althomali, M. R. Atyia et al., “A sys- number of iterations is set to 200. σ �1; that is, only one tematic review of trials investigating the efficacy of exercise optimal solution is retained. Divide the domain into 60 parts training for functional capacity and quality of life in chronic evenly; that is, r � 60. Generally, η can be taken as any kidney disease patients,” International Urology and Ne- number [28] between [10-3, 10], but the value is usually phrology, vol. 32, 2021. small to ensure that the weight correction can truly follow [4] Z. Liu and Y. Liu, “Design of estrus monitoring system for the direction of gradient descent. In this paper, the value is cows based on wechat public platform,” Journal of Chinese 0.003. In the early warning stage of exercise fatigue, rea- Agricultural Mechanization, vol. 12, 2019. sonable measurement units of exercise fatigue time were [5] T. Junji, I. Yudai, I. Torii et al., “A guidance generation divided according to the characteristics of different exercise method based on joint vector model for an exercise training items. It is proved that the Bayesian classification algorithm support system,” IEEJ Transactions on Electronics, Informa- can effectively avoid the sports injury caused by overtraining tion and Systems, vol. 140, no. 3, pp. 364–374, 2020. 6 Journal of Healthcare Engineering [6] S. J. Winser, L. F. Paul, L. K. L. Magnus et al., “Economic evaluation of exercise-based fall prevention programs for people with Parkinson’s disease: a systematic review,” Journal ofAlternativeandComplementaryMedicine(NewYork,N.Y.), vol. 25, no. 12, pp. 1225–1237, 2019. [7] S. C. Howes, D. Charles, K. Pedlow, I. Wilson, D. Holmes, and S. Mcdonough, “User-centred design of an active computer gaming system for strength and balance exercises for older adults,” Journal of Enabling Technologies, vol. 13, no. 2, pp. 101–111, 2019. [8] E. Saraee, Y. Gu, S. Pandit, S. Tran, and M. Betke, “Exer- ciseCheck: data analytics for a remote monitoring and eval- uation platform for home-based physical therapy,” in Proceedings of the 12th ACM International Conference. ACM, Melbourne, Australia, February 2019. [9] C. . Purwanto, Y. Erniyawati, S. Hariyanto, H. I. Muhalla, and E. S. Wijayanti, “,e effect of kegel exercise on the quality of life in post turp patients at Muhammadiyah hospital,” Journal of Vocational Nursing, vol. 2, 2021. [10] M. Efat, S. Rahman, and T. Rahman, “IoT based smart health monitoring system for diabetes patients using neural net- work,” inProceedingsoftheInternationalConferenceonCyber Security & Computer Science (ICONCS), Dhaka, Bangladesh, February 2020. [11] Y. Yu and X. Chi, “Monitoring and management system for college students’ extracurricular physical exercise based on artificial intelligence,”JournalofIntelligentandFuzzySystems, vol. 2, pp. 1–10, 2021. [12] A. Mantelero and M. S. Esposito, “An evidence-based methodology for human rights impact assessment (HRIA) in the development of ai data-intensive systems,” Computer Law and System Review, vol. 41, 2021. [13] S. Jeon and J. Kim, “Effects of augmented-reality-based ex- ercise on muscle parameters, physical performance, and ex- ercise self-efficacy for older adults,” International Journal of Environmental Research and Public Health, vol. 17, no. 9, [14] A. Cavina, E. P. Junior, A. F. Machado, T. M. Biral, and F. M. Vanderlei, “Load monitoring on pilates training: a study protocol for a randomized clinical trial,” Trials, vol. 20, no. 1, [15] T. Zhou, X. Li, and H. Zhao, “Med-ppphis: blockchain-based personal healthcare information system for national physique monitoring and scientific exercise guiding,” Journal of Med- ical Systems, vol. 43, no. 9, pp. 1–23, 2019. [16] C. C. Walton, A. Lampit, C. Boulamatsis, H. Hallock, and M. Valenzuela, “Design and development of the brain training system for the digital maintain your brain dementia pre- vention trial,” JMIR Aging, vol. 2, 2019. [17] F. A. Rathore and A. Afridi, “Is exercise training effective within 12 months of lung resection for non-small cell lung cancer? - a cochrane review summary with commentary,” PM&R, vol. 13, 2021. [18] B. M. Ritter, A. Bynum, M. Gumpertz, and T. L. Butler, “An instructional exercise in gender bias,” Journal of Accounting Education, vol. 54, 2021. [19] S.-Y. Joo, C.-B. Lee, N.-Y. Joo, and C.-R. Kim, “Feasibility and effectiveness of a motion tracking-based online fitness pro- gram for office workers,”Healthcare, vol. 9, no. 5, p. 584, 2021. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Healthcare Engineering Hindawi Publishing Corporation

Design of the Exercise Load Data Monitoring System for Exercise Training Based on the Neural Network

Qin, Panlong; Feng, Wei
Journal of Healthcare Engineering , Volume 2021 – Sep 25, 2021
Download PDF
6 pages

Loading next page...
 
/lp/hindawi-publishing-corporation/design-of-the-exercise-load-data-monitoring-system-for-exercise-z2gDQFMpfi

References (21)

Publisher
Hindawi Publishing Corporation
Copyright
Copyright © 2021 Panlong Qin and Wei Feng. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ISSN
2040-2295
eISSN
2040-2309
DOI
10.1155/2021/7340140
Publisher site
See Article on Publisher Site

Abstract

Hindawi Journal of Healthcare Engineering Volume 2021, Article ID 7340140, 6 pages https://doi.org/10.1155/2021/7340140 Research Article Design of the Exercise Load Data Monitoring System for Exercise Training Based on the Neural Network Panlong Qin and Wei Feng Department of Physical Education, Hebei Academy of Fine Arts, Hebei,050700, China Correspondence should be addressed to Wei Feng; 20122474@stu.nun.edu.cn Received 7 August 2021; Accepted 16 September 2021; Published 25 September 2021 Academic Editor: Balakrishnan Nagaraj Copyright © 2021 Panlong Qin and Wei Feng. ,is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. In order to monitor the sports load data of athletes in sports training, this paper studies the methods and systems of sports load monitoring and fatigue warning based on neural network technology. In this paper, the neural network parallel optimization algorithm based on big data is used to accurately estimate the motion load and intensity according to the determined motion mode and acceleration data, so as to realize the real-time monitoring of the exercise training. ,e results show that the value of η is usually small to ensure that the weight correction can truly follow the direction of the gradient descent. In this paper, 176 samples were extracted from the monitoring data collected by the “National Tennis Team Information Platform,” 160 of which were selected as training samples and the other 16 as test samples. Ant colony size M � 20. ,e minimum value W of the weight min interval is −2, and the maximum value W is 2. ,e maximum number of iterations is set to 200. σ �1; that is, only one optimal max solution is retained. ,e domain is divided into 60 parts evenly; that is, r � 60. Generally, η can be taken as any number [28] between [10-3, 10], but the value is usually small to ensure that the weight correction can truly follow the direction of the gradient descent. In this paper, the value is 0.003. In the early warning stage of exercise fatigue, reasonable measurement units of exercise fatigue time were divided according to the characteristics of different exercise items. It is proved that the Bayesian classification algorithm can effectively avoid the sports injury caused by overtraining by warning the fatigue and preventing the sports injury caused by overtraining. (ACO) is a novel bionic evolutionary algorithm, which 1. Introduction shows excellent performance and great development po- Artificial neural network (ANN) is a kind of nonlinear tential in solving complex optimization problems, espe- dynamic system, which is borrowed from the development cially discrete optimization problems. ACO, as a global of biological neural network [1], the new intelligent in- optimization heuristic algorithm, is used to train the weight formation processing system. With its unique information of neural network, which can avoid the defect of BP neural processing methods, ANN has been widely used in various network. Nowadays, the competition of sports science and fields, such as industrial production monitoring, classifi- technology is increasingly fierce [3]. It is necessary to have cation, prediction and forecasting, analysis and processing, scientific training methods and means in order to improve auxiliary diagnosis of diseases, environmental pollution, sports performance. Sports competition is a big competi- and purification prediction. BP neural network is one of the tion of science and technology. ,e guiding ideology of representative neural network models and has been widely implementing competitive sports science and technology used. However, BP algorithm has some defects, such as low work is to carry out the thought that science and technology efficiency, slow convergence speed, and being easy to fall is the first productive force, carry out the strategy of “in- into local minimum, which restrict the application of BP vigorating sports through science and education,” and network in various fields [2]. Ant Colony Optimization perfect the management system and movement mechanism 2 Journal of Healthcare Engineering monitoring terminal in power transmission, adopts the that sports training and sports science and technology are closely combined [4]. Our research focuses on the actual measures of establishing identification table in the system and judging whether the data frame sent by the terminal needs in training, aiming at the key problems in training practice, to carry out scientific research and public rela- address has been sent before identification, so as to improve tions, and strengthen technical innovation. To promote the the real-time performance of the identification system [14]. scientific sports training, give full play to the leading role of Zhou et al. designed a neural network compensator based science and technology and improve the sports skills of our on the nonlinear error of the motion control system of the athletes and the competitive strength in the world com- image measuring instrument. ,e neural network simu- petition, for our athletes in the 2008 Olympic Games to get lator and compensator of nonlinear servo motor are trained good results to win the gold medal to make contributions by the data of input and output of servo motor. ,e [5]. designed neural network compensator is applied to the high precision positioning system of motion control, which makes the control system show good control performance. 2. Literature Review Simulation results show that the controller is effective [15]. In this paper, the detection data of physiological and bio- With the continuous maturity and development of artificial neural network technology, the intelligent characteristics chemical indexes of national athletes were taken as training samples, and the ACO-BP algorithm was adopted to inte- and capabilities of neural network are increasingly ex- grate ACO and BP algorithm to complete the neural network panded in its application fields. Many problems that cannot training [6, 7]. ,e ant colony neural network prediction be solved by traditional information processing methods model was established for sports load, and the corre- have achieved good results after using neural network, and sponding relationship between physiological and bio- especially in the engineering field it has been widely used. chemical indexes in sports and training load was obtained. With the continuous development of neural network theory itself and related theories and technologies, the Firstly, ACO is used to optimize the neural network weights as a whole to overcome the shortage of BP algorithm which application of artificial neural network will be more in depth and extensive. is easy to fall into local optimum [8, 9]. ,en taking the better weight as the initial value, BP algorithm is used to do In this paper, the test data of Chinese athletes’ physio- further optimization, to overcome the shortcomings of a logical and biochemical indexes are taken as training single ACO training network with long time and low pre- samples, and the ACO and BP algorithm are combined to cision [10–12]. complete the neural network training method, namely ACO- ,e research continues; Jeon and Kim used artificial BP algorithm, to establish the ant colony neural network intelligence and expert system design principle, and the prediction model for the exercise load and obtain the cor- method of simulated medical expert diagnosis, treatment of responding relationship between the physiological and disease thinking process, and compiled computer program biochemical indexes and the training load during the ex- ercise. Firstly, ACO is used to optimize the weight of the can help doctors to solve complex medical problem, as the doctor concluded that it is an important auxiliary tool for neural network globally to overcome the shortcoming of BP algorithm which is easy to fall into local optimal. ,en, with the treatment of disease and prognosis. Medical expert system is an important application of artificial intelligence the optimal weight as the initial value, BP algorithm is used in medicine. It is a collection of knowledge, computer to do further optimization, so as to overcome the disad- technology, network technology, communication tech- vantages of a single ACO training network such as long time nology, database technology, and medical science. ,e and low accuracy. development of medical expert system has become an important topic in modern medicine [13]. Cavina et al. 3. Research Methods designed a multiprotocol recognition system based on BP artificial neural network. ,e specific research contents of ,eoretical analysis proves that a feedforward net with a this paper include the design of multiprotocol recognition single hidden layer can map all continuous functions only eigenvalue, the data of a certain length of frame head is when learning. Two hidden layers are required when the taken as the eigencode, and the eigenvalue is obtained after function is discontinuous. Increasing the number of hidden normalized processing. ,e selection of multiprotocol layers can improve the nonlinear mapping ability of BP learning algorithm, simulation analysis, and comparison of network, but when the number of hidden layers exceeds a several BP neural network algorithm is according to the certain value, the performance of BP network will decline. In convergence rate and recognition rate of conjugate gradient 1998, Hecht-Nielson proved that a continuous function in descent method as the multiprotocol recognition system BP any closed interval can be approximated by a BP network neural network identification algorithm. As per the es- with a hidden layer. ,erefore, this paper will adopt a three- tablishment of multiprotocol recognition network, through layer BP network with a hidden layer, namely, R-N-1. Neural simulation, according to the recognition rate, the maxi- network structure model is where n is the number of input mum number of hidden layer of BP neural network suitable nodes, r is the number of hidden nodes, and the number of for the multiprotocol recognition system in this paper is 9. output nodes is 1. ,e number of neurons in the input layer ,e design of multiprotocol identification system, aiming of the neural network is determined by the influencing at the characteristic of the relatively fixed address of the factors [16, 17]. In this paper, the physiological and Journal of Healthcare Engineering 3 biochemical indexes which can obviously reflect the phys- 1 l iological function and the professional training years are taken as the influencing factors. ,e number of nodes in the output layer is set as 1, and the value range is [−1, 1], which is 2 1 used as the quantified value to measure the load. ,e closer the value is to 1, the smaller the load is, and vice versa. ,e role of hidden layer nodes is to extract and store the inherent rules from the samples. Each hidden layer node has several weights, and each weight is a parameter to enhance the mapping ability of the network. If the number of hidden Figure 1: BP network structure. nodes is too small, the network’s ability to obtain infor- mation from the sample is poor, which is not enough to normalized input and target data are subject to normal summarize and reflect the sample rule of the training distribution; i.e., sample. If the number of hidden nodes is too large, it may also learn to remember the irregular content of the sample, [pn, meanp, stdp] � prestd(p), (2) such as noise, which reduces the generalization ability. where pn is the input direction after vectorization, p is a Generally, the following empirical formula can be used group of collected data (vector), and meanp is the input for the number of hidden layer nodes in three-layer forward direction after vectorization. Enter the mean value of the network: vector, and STDP is the deviation of the input vector. ,e √���� � algorithm executed is m � n + 1 + a, (p − meanp) m � log 2 , (1) pn � . (3) √�� stdp m � nl. As physiological and biochemical indexes, some indexes M is the number of nodes in the hidden layer, N is the are correlated. ,rough the principal component analysis of number of nodes in the input layer,L is the number of nodes standardized data, the redundant components of sample in the output layer, and α is a constant between 1 and 10. In data can be eliminated; that is, the components that con- this paper, the number of hidden layer nodes is determined tribute little to the overall change of the data set, so as to according to equation (1). To sum up, the BP neural network reduce the data dimension. ,e processing function of adopted in this paper is shown in Figure 1. MATLAB is [Ptrans, TransMat] � prepca(pn, 0.02); (4) 3.1. Processing of Sample Data. Tennis sport has the longer where 0.02, the second parameter in the prepca function, game time, the intensity of the intermittent type, short smell, indicates that the main element is an element whose square and explosive force action composition. Tennis is more and deviation from the mean is greater than 2% of the variance of more towards the strength and speed against the direction of the vector. TransMat is a change matrix with major elements. development; the physical requirements are very high. With In this paper, the blood data and urine data were ana- the increase of professional training years, various functions lyzed by principal component analysis. ,rough the analysis of the body will change, and the ability to adapt to the load and processing of a group of processed blood test data and will also change. Athletes with short training years can bear urine data of national tennis team players, the urine sample less amount of exercise than those with long training years. data were changed from the original 9. ,e dimension (8 ,e same amount of exercise may be a large exercise load for dimensions of urine index and professional training years) young players. ,erefore, the professional training years can was reduced to 6, and the data of blood data were reduced to be used as an important influencing factor of load fore- 13 from 15 (14 dimensions of blood index and professional casting to improve the accuracy of assessment ability. training years). ,e training samples used in the experiment are from the tennis team sports collected by the “National Tennis Team Information Platform” physiological and biochemical in- 3.2. Basic Steps. In the case that the structure of the neural dexes monitoring information of the mobilized competition network has been determined, the basic steps of training the and daily training. Due to the size of the original test data, in neural network by the ACO-BP method are as follows: order to avoid the difference of physical meaning and unit of Step 1: initialize: input vector on the network model, the difference of bit is obvious. Before the training of neural network, the training ,e weights of all the weights of the interval [W and min samples need to be processed in advance, that is, normalized W ] are evenly divided intor subregions. Each region max processing [18, 19]. In order to train the function better, we that the points on the boundary is the value of an also can avoid the network failure caused by being too large. alternative for each parameter setting up a list of ,e numbers are collected in this article. ,e data are pheromones as shown in Table 1. At the initial moment, mapped between [−1, 1] and normalized, so that the set each point to have the same amount of pheromone 4 Journal of Healthcare Engineering Table 1: Pheromone table of weights. Step 7: the neural network is further trained by BP algorithm, the best weights of σ group found by ANT Label 1 2 ... r + 1 Divide the scale a1 a2 ... ar + 1 colony algorithm are taken as the initial weights of BP ,e pheromone values (1) (2) ... (r + 1) algorithm, the error between network output and actual output is calculated, and the error is propagated one- way from the output layer to the input layer to adjust tau 0, pheromone volatilization coefficients rho, and the weights. Repeat the process until the stop condi- time t, and the number of loop NC is zero. Set the tions are met. maximum number of loops of the ending condition, Step 8: use validation samples to test the generalization MAXNC. ability of the trained neural network. If the validation Step 2: release all M ants. Ant K uses the following error meets the requirements, quit the program. probability formula as a path selection rule. ,at is, Otherwise, go to Step 1 and start training again. select an element in each set Ipi(1≤I≤m). 4. Result Analysis τ(i) p (i) � . (5) In order to verify the network model, the author has done a 􏽐 τ(i) 1≤j≤m lot of computer simulation experiments. In women’s team sports, as an example, the blood index data of the team To record the label of the point passed by the ant, a members were collected from the monitoring data collected value is selected for the weight and recorded in K. After by the “National Tennis Team Information Platform.” 176 the ant selects the values for all the weight parameters, samples were taken, and 160 of them were selected as the ant completes a traversal, and all the values it training samples and the other 16 were selected as test chooses to record constitute all the parameters of the samples. Ant colony size M � 20. ,e minimum value of the neural network. weight interval W is −2, and the maximum value W is min max Step 3: repeat Step 2 until all the ants reach the food 2. ,e maximum number of iterations is set to 200. σ �1; that source. is, only one optimal solution is retained. ,e domain is Step 4: let T← +M, NC← NC + 1. Use the weight evenly divided into 60 parts; that is, r � 60; η can generally parameters selected by each ant to calculate the take any number between [10-3, 10], but it usually takes a output results of the neural network and the error E, small value to ensure that the weight correction can really record the weight of σ group with small error and the descend along the direction of the gradient. In this paper, the current optimal solution Emin, and compare the size value is 0.003. ,e specific parameter settings of the network of Emin and E0. If Emin≤E0, go to Step 8; otherwise, are shown in Table 2. go to Step 5. According to the actual situation of the national tennis Step 5: update the pheromone of each element team and the need of preparing for the Olympic Games, the according to the pheromone regulation rules below. national tennis team information platform is designed and Pheromone regulation: as time goes by, the phero- developed with Delphi7.0 development tool and SQL mone that was left behind gradually fades away. ,e Server2000 database. ,e information platform includes volatility coefficient ρ represents the persistence of the athletes’ basic information, athletes’ training and competi- pheromone, and 1−ρ represents the vanishing degree tion information data, image and video analysis and training of the pheromone. After n time units, the ants travel monitoring data, which is a set of analysis and management from the nest to the food source, and the pheromones system for tennis. It mainly stores the basic information of along each path are adjusted according to the fol- Chinese tennis players, athletes’ injuries and rehabilitation, lowing formula: nutritional recovery information, athletes’ special energy, physiology and biochemistry, training load, training τ 􏼐I 􏼑(t + m) � pτ I 􏼁 (t) + Δτ 􏼐I 􏼑, j pi j Pi j pi method, training course content, training effect, competition process, competition results, and other training match in- ⎪ formation data. From the perspective of function, it mainly (6) k e includes data management, video image management, and Δτ I � 􏼐 􏼑 j pi data analysis. 0, ,e output results corresponding to the samples were jointly determined by experts, team doctors, and other where Q is a constant, used to adjust the pheromone people in combination with the corresponding training adjustment speed. e is the output error when a set of plans and athletes’ performance at that time. We use minus 1 weights selected by the KTH ant is taken as the weights for body function. Poor athletes cannot adapt to the current of the neural network and is defined as e � |o − o |, training plan and should immediately adjust the amount of where o ando are the actual and expected outputs of training. Let us denote the athlete by 0. Being able to adapt to the neural network. As you can see, the smaller the the current training plan, exercise amount is appropriate and error, the larger the corresponding pheromone can maintain the training plan for training; 1 indicates that increase. the body function is in a completely normal state, the Journal of Healthcare Engineering 5 Table 2: ACO-BP and BPNN parameter settings. ,e name of the M W W Q r NACO NBP EO min max ACO-BP 20 −2 2 0.005 60 200 0.003 12000 0.005 BPNN −0.1 0.1 0.003 20000 0.005 Table 3: Comparison of simulation test results. ,e mean square error of the Nth iteration ,e experimental method N � 100 N � 500 N � 1000 Standard BP algorithm 0.00977462 0.0063423 0.00429856 BP algorithm for adding the momentum term 0.00868283 0.00416346 0.000683606 ACO-BP algorithm 0.0071246 0.00198015 8.9427e − 5 athletes can fully adapt to the current amount of exercise, by warning the fatigue and preventing the sports injury and can appropriately increase the amount of training to caused by overtraining. ,e research of this paper is a improve the training effect. ,e test results obtained through preliminary study, combined with ability and the limitation the simulation test are listed in Table 3. of technical conditions. ,ere are a lot of improvement, From the above experimental results, the following mainly having the following several aspects: (1) the ant conclusions can be drawn: the three-layer front is based on colony algorithm is a new bionic algorithm. Now still not ant colony optimization algorithm presented in this paper. mature, the parameters of the ant colony algorithm selection Based on the training model of feedback neural network (BP theoretical guidance remain to be further strengthened, neural network), a new network training algorithm is because it directly relates to the parameter selection of established. ,e experimental results are analyzed and application effect of the algorithm. (2) ,e ant colony al- compared with the standard BP algorithm and the improved gorithm is further improved to increase the convergence of BP algorithm with momentum term, ACO-BP. ,e algo- the improved ant colony algorithm to meet the needs of rithm has a fast convergence speed and can achieve a smaller large-scale network. (3) Integrate qualitative and quantita- value of mean square error. ,erefore, the convergence tive problems, and quantify fuzzy concepts, so as to better process of this method has obvious advantages and stability. meet the needs of users. 5. Summary of This Paper Data Availability In order to monitor the sports load data of athletes in sports ,e data used to support the findings of this study are training, this paper studies the methods and systems of available from the corresponding author upon request. sports load monitoring and fatigue warning based on neural network technology. In this paper, the neural network Conflicts of Interest parallel optimization algorithm based on big data is used to accurately estimate the motion load and intensity according ,e authors declare that they have no conflicts of interest. to the determined motion mode and acceleration data, so as to realize the real-time monitoring of the exercise training. References ,e results show that the value of η is usually small to ensure that the weight correction can truly follow the direction of [1] M. A. Yan, H. Deng, and X. J. University, “Design of sports gradient descent. In this paper, 176 samples were extracted training correction system based on virtual reality,” Infor- from the monitoring data collected by the “National Tennis mation & Technology, vol. 28, 2019. Team Information Platform”, 160 of which were selected as [2] Y. Lin, Q. Qu, Y. Lin et al., “Customizing robot-assisted training samples and the other 16 as test samples. Ant colony passive neurorehabilitation exercise based on teaching sizeM � 20. ,e minimum value W of the weight interval min training mechanism,” BioMed Research International, is −2, and the maximum value W is 2. ,e maximum vol. 2021, no. 31, 10 pages, Article ID 9972560, 2021. max [3] A. A. Ibrahim, O. W. Althomali, M. R. Atyia et al., “A sys- number of iterations is set to 200. σ �1; that is, only one tematic review of trials investigating the efficacy of exercise optimal solution is retained. Divide the domain into 60 parts training for functional capacity and quality of life in chronic evenly; that is, r � 60. Generally, η can be taken as any kidney disease patients,” International Urology and Ne- number [28] between [10-3, 10], but the value is usually phrology, vol. 32, 2021. small to ensure that the weight correction can truly follow [4] Z. Liu and Y. Liu, “Design of estrus monitoring system for the direction of gradient descent. In this paper, the value is cows based on wechat public platform,” Journal of Chinese 0.003. In the early warning stage of exercise fatigue, rea- Agricultural Mechanization, vol. 12, 2019. sonable measurement units of exercise fatigue time were [5] T. Junji, I. Yudai, I. Torii et al., “A guidance generation divided according to the characteristics of different exercise method based on joint vector model for an exercise training items. It is proved that the Bayesian classification algorithm support system,” IEEJ Transactions on Electronics, Informa- can effectively avoid the sports injury caused by overtraining tion and Systems, vol. 140, no. 3, pp. 364–374, 2020. 6 Journal of Healthcare Engineering [6] S. J. Winser, L. F. Paul, L. K. L. Magnus et al., “Economic evaluation of exercise-based fall prevention programs for people with Parkinson’s disease: a systematic review,” Journal ofAlternativeandComplementaryMedicine(NewYork,N.Y.), vol. 25, no. 12, pp. 1225–1237, 2019. [7] S. C. Howes, D. Charles, K. Pedlow, I. Wilson, D. Holmes, and S. Mcdonough, “User-centred design of an active computer gaming system for strength and balance exercises for older adults,” Journal of Enabling Technologies, vol. 13, no. 2, pp. 101–111, 2019. [8] E. Saraee, Y. Gu, S. Pandit, S. Tran, and M. Betke, “Exer- ciseCheck: data analytics for a remote monitoring and eval- uation platform for home-based physical therapy,” in Proceedings of the 12th ACM International Conference. ACM, Melbourne, Australia, February 2019. [9] C. . Purwanto, Y. Erniyawati, S. Hariyanto, H. I. Muhalla, and E. S. Wijayanti, “,e effect of kegel exercise on the quality of life in post turp patients at Muhammadiyah hospital,” Journal of Vocational Nursing, vol. 2, 2021. [10] M. Efat, S. Rahman, and T. Rahman, “IoT based smart health monitoring system for diabetes patients using neural net- work,” inProceedingsoftheInternationalConferenceonCyber Security & Computer Science (ICONCS), Dhaka, Bangladesh, February 2020. [11] Y. Yu and X. Chi, “Monitoring and management system for college students’ extracurricular physical exercise based on artificial intelligence,”JournalofIntelligentandFuzzySystems, vol. 2, pp. 1–10, 2021. [12] A. Mantelero and M. S. Esposito, “An evidence-based methodology for human rights impact assessment (HRIA) in the development of ai data-intensive systems,” Computer Law and System Review, vol. 41, 2021. [13] S. Jeon and J. Kim, “Effects of augmented-reality-based ex- ercise on muscle parameters, physical performance, and ex- ercise self-efficacy for older adults,” International Journal of Environmental Research and Public Health, vol. 17, no. 9, [14] A. Cavina, E. P. Junior, A. F. Machado, T. M. Biral, and F. M. Vanderlei, “Load monitoring on pilates training: a study protocol for a randomized clinical trial,” Trials, vol. 20, no. 1, [15] T. Zhou, X. Li, and H. Zhao, “Med-ppphis: blockchain-based personal healthcare information system for national physique monitoring and scientific exercise guiding,” Journal of Med- ical Systems, vol. 43, no. 9, pp. 1–23, 2019. [16] C. C. Walton, A. Lampit, C. Boulamatsis, H. Hallock, and M. Valenzuela, “Design and development of the brain training system for the digital maintain your brain dementia pre- vention trial,” JMIR Aging, vol. 2, 2019. [17] F. A. Rathore and A. Afridi, “Is exercise training effective within 12 months of lung resection for non-small cell lung cancer? - a cochrane review summary with commentary,” PM&R, vol. 13, 2021. [18] B. M. Ritter, A. Bynum, M. Gumpertz, and T. L. Butler, “An instructional exercise in gender bias,” Journal of Accounting Education, vol. 54, 2021. [19] S.-Y. Joo, C.-B. Lee, N.-Y. Joo, and C.-R. Kim, “Feasibility and effectiveness of a motion tracking-based online fitness pro- gram for office workers,”Healthcare, vol. 9, no. 5, p. 584, 2021.

Journal

Journal of Healthcare Engineering – Hindawi Publishing Corporation

Published: Sep 25, 2021

There are no references for this article.