Application of BP neural network algorithm in visualization system of sports training management

When constructing the algorithm model of sports training action classification, the accuracy of action classification has an important impact on the algorithm model. How to improve the algorithm model to improve the accuracy of sports training action classification needs further research. Based on BP neural network algorithm, this paper carries out the modeling of BP neural network signal classification algorithm and the construction of BP neural network, and deduces the BP algorithm in detail. Firstly, this paper applies genetic algorithm to the initial parameter selection of BP algorithm to avoid the local optimization problem. When carrying out chromosome coding, binary coding is easy to cause the problem of too long coding string, which also needs to be restored and decoded. The algorithm runs too long and the learning accuracy is not high. Therefore, this paper uses real coding. Through simulation analysis, it can be seen that the classification accuracy of the improved algorithm model is significantly higher than that of the simple BP algorithm. In addition, this paper analyzes the requirements of the sports training management visualization system, introduces the system structure framework and network topology, describes in detail the functions of the user information management module, the training plan management module, the training test management module, the competition information management module and the scientific research information management module, and tests the visualization function of the system. Finally, this paper analyzes the problems existing in the current sports training management, and puts forward the development strategy of sports training management based on this, which lays a theoretical foundation for the scientific development of sports training.


Introduction
BP neural network has been widely used in many fields. It uses backpropagation algorithm, which is also the most mature algorithm in neural network research. It has been widely used in image recognition and processing, data information recognition and processing, intelligent pattern recognition and control, enterprise management, market analysis and other fields (Ding et al. 2011). At present, BP neural network is widely used in economic, military and chemical fields, and its application advantages and effects are very significant (Cui and Jing 2019). Therefore, in the period of highly developed information technology and economic development, it is of theoretical value to study the application of BP neural network in sports training management (Guo et al. 2011). Visualization technology has also been widely used in various industries to build a three-dimensional virtual framework, which can present data in a three-dimensional state and realize data interactive processing (Guo et al. 2017). At present, visualization technology is widely used in medical and health information. People pay more and more attention to the interactivity of health data. Visualization technology displays users' own health information in three-dimensional form, and users can intuitively and dynamically understand health data. Relying on visualization technology, sports training management can help coaches and athletes more clearly analyze training results, scientifically and reasonably set up sports plans, and improve the efficiency of sports training management (Cheng and Teizer 2013). Therefore, it is of practical value to study the application of visualization technology in sports training management. Physical training in Colleges and universities is very common, but the management mode is relatively backward (Nurbekova et al. 2020). The purpose of physical training in Colleges and universities is to help students get healthy physique, as well as obtain certain honors through competitive sports. In terms of talent training of students, they pay too much attention to training results, have a weak concept of physical education, and lack a comprehensive and scientific physical training management system (Daniłowicz-Szymanowicz et al. 2013). Based on the idea of BP neural network, this paper constructs a classification algorithm model of sports training actions, analyzes the requirements of the current sports training management visualization system, designs a sports training management visualization system including user information management, training plan management, training test management, competition information management and scientific research information management, and analyzes the problems existing in the current college sports training management (Yuan et al. 2020;Alahakone and Senanayake 2010). This paper discusses the development strategy of sports training management in the new period, which lays a theoretical foundation for improving the level of sports training management.

Related work
Neural networks are widely used in various fields. According to literature statistics, BP neural networks account for more than 80% of all neural networks . This paper presents a method to improve the performance of BP neural network by using multiple parameters to adjust the activation function. The literature introduces that activation functions can realize complex mapping functions. An activation function is constructed to improve BP neural network and optimize the traditional transfer mode (Yang et al. 2019;Zhang et al. 2023). In order to further reduce the saturation region, the activation function is adjusted, the number of model training is reduced, and the convergence speed is significantly improved. In the literature, BP neural network algorithm and de algorithm are combined to optimize the weight of neural network, so as to improve the convergence speed (Cui and Qin 2018). In the literature, PSO and other techniques are applied to the training of BP neural network algorithm, which can significantly improve the convergence speed and fitting accuracy. At present, there is a ''one pot'' phenomenon in sports training management. Athletes lack targeted and professional training, especially in the classification of sports movements, there is a lack of intelligent equipment (Dugalić 2011). The literature points out that the current athletes' training record management is extensive, only using the method of stage record, and lack of information management. For coaches and athletes, the management system with neural network learning technology and visualization technology can improve management efficiency, and achieve scientific and reasonable training (Muhammad and Halim 2016). Using neural network algorithm to build a motion classification model can help athletes carry out personalized training and further improve their professional level (Xia et al. 2019;Sangaiah et al. 2023;Khanduzi and Sangaiah 2023). Relying on information technology to build a sports training management system, to manage athletes' personal information and training data, to assist coaches in formulating training plans and competition information, and to establish a communication channel between parents and schools. Visualization technology has been widely used in military, teaching, medical, financial, chemical industry, architecture and other fields, especially in medicine, which has shown remarkable results (Lin et al. 2019). Integration and processing of medical information through visualization technology can realize the organic fusion and data analysis of various medical information. Visualization technology has an important application in the field of human body planning. Relying on the mobile phone terminal, the literature has built a platform for recording and visualizing human daily activities (Wong et al. 2014). The platform shows the virtual scene of human daily activities. The sensors on the mobile phone can judge the status of human activities, such as sitting, standing, walking and running, and map the action information to the virtual model of the platform to realize the visualization of human daily actions. The mapping method can be ordinary mapping or specific mapping. In addition to realizing the consistency between the human virtual model of the platform and the real human activities, the real human actions can also be mapped to other actions in the virtual world according to the special requirements of users (Michalakis et al. 2018). Visualization technology can be applied in sports training, and visually analyze athletes' movements and techniques. In addition to the visual system, the requirements of sports training for athletes and coaches also need to be further improved. Training of athletes should be combined with physical education and sports training. The growth of athletes needs both education and technical training. At present, the task of athletes' training undertaken by colleges and universities is more about the improvement of athletes' technology, and there is a lack of cultural level and scientific literacy. According to the literature, with the improvement of athletes' technical level, personal cultural literacy should also be improved. It is necessary to closely combine skill training and cultural education to achieve the all-round development of athletes.

BP neural network structure
The structure of BP neural network is divided into three layers, and the specific topology is shown in Fig. 1.
Suppose there are n neurons in the hidden layer and m neurons in the output layer. The output of each neuron is: It can be seen from the above formula that the mapping from n-dimensional space vector to m-dimensional space vector is completed. The activation function f (x) in the data model is unipolar sigmoid function: As can be seen from the above formula, f (x) shows the continuous differentiable feature, and also shows the following features: According to the actual application needs, the bipolar sigmoid function can also be used: The weight and threshold learning method of BP neural network adopts BP algorithm. The learning process of the neural network is divided into two parts. The first part is the forward propagation of the signal and the other part is the backpropagation of the error. The propagation direction of the learning signal and the error signal is opposite. Through learning and training, the output value is continuously adjusted and optimized. Then, calculate the output value generated after the input of the p-th sample, and finally obtain the mean square error value according to the minimum mean square error criterion.
The mean square error shall be the sum of the square errors of each output unit: After all learning samples are input into BP neural network, the total error can be obtained as follows: According to the gradient descent method, the weight correction obtained is: For the output layer: Application of BP neural network algorithm in visualization system of sports training… 6847 where: So there are: Of which: For intermediate hidden layers: where: So there are: The steps of BP algorithm are: Step 1 Weight initialization; Step 2 Input in sequence P learning samples. Set the current input as P samples; Step 3 Calculate the output of each layer in turn; Step 4 Calculate the backpropagation error of each layer: Step 5 Analyze the trained sample P. if p \ p, return to step 2 to continue the calculation; If p = p, proceed to the next step; Step 6 The weights and thresholds of each layer of BP neural network are trained and modified according to the weight correction formula; Step 7 Recalculate the output value according to the new weight. If the maximum number of learning times is reached, the learning will be terminated. Otherwise, go to the second part to continue a new round of learning.

Algorithm improvement
Gradient descent method is a calculation method adopted by BP algorithm. This method has the problem of poor global search effect, but strong local search effect. Genetic algorithm adopts the global optimization method, which has the advantage of strong global search effect, but the local search effect is poor. Therefore, in order to improve the calculation effect of BP algorithm, this paper combines genetic algorithm with BP algorithm to realize complementary advantages.
The learning process of BP neural network is mainly to continuously optimize the weights and thresholds of the learning network. If these two parameters are incorrectly selected, BP algorithm will obtain local optimal results. In this paper, genetic algorithm is applied to the initial parameter selection of BP algorithm to avoid the local optimization problem. In chromosome coding, binary coding is easy to cause the problem that the coding string is too long, and it also needs to be restored and decoded.
There are some problems such as too long running time and low learning accuracy.
In this paper, real number coding is used in the following form: Define the network error function: The objective function value is inversely proportional to the fitness value, so the function formula is as follows: Suppose the population size is m, and the probability of individual I being selected is p: According to the explanation of the above article, this paper adopts the real number coding method, and the new individual: The mutation operator adopts the uniform mutation strategy, and the new gene values with multiple mutation points are:

Simulation analysis
This paper uses SPSS software to process the data of the input samples through factor analysis. Through the analysis, we know that the kmo value is 0.779, which is significantly greater than 0.5. Therefore, this experiment can be used for factor analysis. After the data analysis of the experimental samples, the input value changes from 13 to 3 dimensions. Then the Matlab toolbox is used to program and train the network with 6 hidden layer nodes and 8 hidden layer nodes, respectively. The test samples are used to test the performance of the network. Table 1 shows the results of hidden node 6. After analysis, the result of hidden node 8 is shown in Table 2.
4 Physical training management visualization system

System requirements analysis
For users' needs for basic information management. The main part of the system covers the principal in charge, director, physical education teacher and special students, so a special account and password are set for each user as the login key. Among them, the most detailed user information belongs to the information management system for sports students. The system includes their name, gender, projects, sports experience and other basic information. It is also an important part of basic information management. First of all, the need to improve training plan management should include time period training plans, such as weekly training plans. The scientific and reasonable training plan is the best choice to train excellent students and the only way to improve their project performance. A good training plan has a guiding role and can avoid most useless training, which shows the importance of a scientific training plan. Therefore, the sports training plan will be regarded as one of the most important structural units when designing the training management information system for sports specialty students.
Taking the college sports specialty students as an example, in view of their need to complete different training tasks and objectives at this stage, when formulating the annual plan, it is necessary to scientifically arrange and combine the time period, tasks, techniques, equipment, etc. based on the national standard outline and in combination with the training plan and the situation of the specialty students. Generally speaking, the formulation of the annual plan is closely related to the competition cycle. Generally, a year is divided into three periods according to the competition schedule, namely, the training period, the competition period and the recovery period. In addition, the established plan should be adjusted and supplemented according to the actual training situation and competitive needs. Compared with the long-term training, the specific training content of each relatively short course is determined by the class hour plan, and it is the most reasonable way to scientifically formulate the training process, content and load intensity in line with the characteristics of specialty students. Therefore, when building the plan management module, we should focus on the requirements of on-demand query and adjustment.
Secondly, the needs of training effect evaluation. The training efficiency represents whether the physical education students' training is scientific and reasonable, and whether they can reap good benefits. According to this standard, the training can be made reasonable changes. Among them, physical fitness, technology and competitive psychology can be used as the sub contents of evaluation, and the evaluation methods of training effect can be divided into subjective and objective. The subjective evaluation includes the self-evaluation of sports students and the subjective evaluation of teachers. Special students expound their training experience in the form of self-evaluation, and carry out more subjective evaluation. They can also standardize their self-evaluation by relying on the Self-evaluation Scale; Physical education teachers can make judgments through professional observation, and make periodic evaluation and summary on the standards of  Application of BP neural network algorithm in visualization system of sports training… 6849 physical education students, training effects, attitudes and psychology. The objective evaluation is completed with the help of the index system, and the big data analysis system is used to objectively evaluate the training effect. Therefore, relying on the scientific sports training management system, it is very important to evaluate the athletes comprehensively and scientifically. For coaches, it can avoid the problem of wrong judgment or evaluation due to personal experience. For athletes, it can be adjusted according to the scientific training evaluation results given by the system. In addition, the demand for competition information management. Training and competition promote each other. Competition is both the result and the test of training. Therefore, it is necessary to list the competitive management separately in the specialized student training management information system. The competitive information should include the competition time, place, project category, contestants and results. The accumulation of longterm competition data and the data recording of the recent status of the students with special skills can not only be used as a reference for selecting the students with special skills, but also be more conducive to the teachers' rational arrangement of tactics, so as to achieve better results. In addition, the competition is an extension of training. The digital management of the competition schedule can not only assist teachers to summarize the lack of training, but also point out the direction of training for special students. Therefore, the competition information management module needs more attention.
Finally, for the demand of scientific research information management, the combination of practice and scientific research is the only way to improve the training level of special students. Because the scientific research information management involves many disciplines, the system is complex, and the data are miscellaneous, the traditional recording method is no longer suitable for today's sports training requirements, and the previous data analysis and technical methods have some drawbacks. Therefore, it is very necessary to choose to store and integrate the relevant data in physical education students' training as the basic function of scientific research management. The traditional recording method is not only time-consuming and laborious, but also unable to make a comprehensive analysis. The sports training management system can give clear and dynamic data information, not only can scientifically analyze the trend changes, but also can carry out targeted analysis and evaluation for a certain skill of a specific athlete.

System architecture design
Through the analysis and demonstration of requirements, the five functional modules of the system are finally determined. The system structure framework is shown in Fig. 2.
The basic hardware structure of the system adopts the b/s architecture mode. With the server as the node, it is expanded into several parts: athlete management, coach management, physical trainer management and background management. The topology of the system is shown in Fig. 3.

System function module design
The functions of user information management module include system user management and basic information management of special students. The system user management is the entrance set for the three types of users: principals in charge, directors of teaching and research offices and sports coaches. Users can edit user name, password, department and login. The account number of sports specialty students is created by the above three types of users. Specialty students can use their own account to log in to the system, and the file data under the account can be used as an important reference for selecting students and determining options in future.
Secondly, the training plan management module is a module for physical education teachers to add, modify and query training plans, covering three types of plans: year, week and class hour. For physical education teachers, the application of this module can test and feedback the training plan during the training process. Under this module, edit the categories, training objects, objectives, time, contents and notes of the annual, weekly and class hour training plans.
The training and test management module mainly includes two sub modules: score management and test summary list. The former is used by coaches to set up test categories and indicators for physical education students. This module can help teachers quickly set targeted test content for students, and improve test efficiency and effect. The test summary module is the teacher's evaluation of students' status, which is entered into the system as the basis for evaluating training effectiveness. This module can establish a communication channel between teachers and students, and help students better improve their professional skills through correct evaluation.
The competition information management module includes two parts: competition management and competition technology statistics. The former aims to input and store information related to competitive events, and the latter is used to summarize and analyze the technical information in the competition. The function of scientific research information management module is to store, adjust and analyze the biochemical indicators of physical education students. In this module, you can query the database information of physical education students, generate model curves and data change trends.

System test
In the system test, several common movements in daily physical training are selected, including standing, sitting, walking and lying down. The joint posture angles of the above four movements are collected. The detection positions are mainly trunk, left thigh, right thigh, left calf and right calf.
A total of 60 groups of motion samples were collected for the experiment, of which 36 groups were randomly selected for data training and 24 groups for data testing. Each group of motion data includes three-dimensional attitude angles of five joints. Feature extraction is carried out for each group of motion, including three feature information of time-domain feature mean, standard deviation and variance, which is expanded to 45 dimensions, and then the first dimension feature is added, and the action category identifier is added.
After training, the action classification test results of BP neural network are shown in Fig. 4.
The error of motion classification is shown in Fig. 5. For the classification of sitting posture, two data were misjudged as lying, while for walking, two were misjudged as sitting posture, and one was misjudged as upright. Standing, sitting and lying down belong to static posture, with relatively fixed angle, small change and high recognition rate; Walking belongs to action posture, and the recognition rate is low. The comparison shows that the extraction and recognition rate of dynamic action features needs to be improved. As a prominent function of the system, data visualization enables users to view the training differences, quality differences and temporal changes of physiological data of athletes through the comparison function.
Through the histogram, the differences of training results of different athletes can be visually compared. As shown in Fig. 6, the average speed and average power of athlete 2 in training are better than the other two.
The trend of physiological and biochemical data of athletes over a period of time can be seen by comparing the broken line chart. The physiological and biochemical data of athlete I is shown in Fig. 7.

Analysis of existing problems
There are some problems in the research of sports training management. First, the training management goal positioning may conflict with the athlete goal positioning. According to the survey, some sports teams have formulated some special policies to allow athletes to participate in training, such as exemption from some subject courses and extra points for courses, which has changed the original intention of training. The purpose of some sports team training management is to win honor for the collective through competitive performance. The excessive pursuit of performance and reputation makes the goal of sports team training management too utilitarian and realistic, which deviates from the fundamental of sports competition. Therefore, whether the training target positioning is consistent with the athletes is the main problem at present.
Second, the level of trainers is uneven. Some ordinary school trainers have rich theoretical knowledge, but lack sports training experience, while professional and experienced coaches lack deep scientific research ability. The coach training system has not been formed, the coaches have few opportunities to learn, the lack of teaching energy and other problems have led to the deep-rooted traditional three-level training network, which has greatly hindered the development and expansion of the sports team.
Thirdly, the contradiction between athletes' learning and training is still prominent. A large number of training that is not completely scientific leads to the dispersion of athletes' energy, the knowledge and cultural level is not easy to meet the national requirements, and finally makes the comprehensive quality of athletes in our city different. Secondly, the conflict between the schedule and the curriculum has affected the learning of professional courses. In the long run, athletes tend to have evasive psychology, and skipping classes and training has become a habit,    Fig. 7 Line chart of physiological and biochemical data changes of athletes which is extremely detrimental to the formation of athletes' healthy mentality and team management. Fourth, the team management conditions are not perfect. Some colleges and universities are difficult to ensure the material security of athletes in the case of lack of funds. However, the training rules, incentive policies and reward and punishment system of coaches are still not perfect. Some coaches lack institutional awareness, and it seems to be a common phenomenon that they only pay attention to training and despise the achievement of training goals.
Fifth, the feedback of training management supervision and evaluation is unreasonable. Without the help of intelligent sports training management system, coaches usually choose simple physiological indicators according to their inherent sports experience, so it is not scientific. Most athletes cannot find and deal with sports injuries in time, which affects the implementation of the whole training plan.

Development strategy research
First, schools should strengthen their understanding of the overall goals and training objectives, and be good at using intelligent systems to enhance the training benefits of athletes. The forms of physical training in Colleges and universities should be innovated and the concepts should keep pace with the times. We should strengthen the macro control of physical training, and formulate a management, evaluation, reward and punishment system in line with the characteristics of the school. Only in this way can the school physical training develop in a good direction. When colleges and universities carry out physical training, the form should not be confined to the traditional model, but should improve their cognition and change their thinking. Through physical education, sports activities, competition and debate, theoretical training and other ways, covering the education and teaching related to physical education from multiple angles, and striving to build physical training as a routine and daily teaching mode.
Second, strengthen the team building of coaches and improve their professional level and ability. Coaches are the leading figures in training, and their level determines the professional development and height of athletes. Good work style, lofty professional ideal and careful training plan ability all directly affect the enthusiasm of trainers. Coaches with digital information literacy can improve the professional skills of athletes with the help of intelligent sports training management system. In addition, the sense of responsibility and justice is an inevitable requirement to win the trust and dependence of athletes and enhance the credibility of coaches. Only with the above requirements can we promote the healthy development of college sports training.
Third, improve the professional standards of athletes and expand the source of students. We can cultivate athletes' interest in sports and stimulate their enthusiasm for sports by strengthening their learning of sports technology theory, sports technology, sports nutrition, sports fatigue recovery and other knowledge. And the athletes who master more sports theory knowledge are more capable of solving sudden problems in training. Athletes should also have a strong information level, improve their information technology level, be able to skillfully use the intelligent sports training system, master certain sports training and computer technology, and realize scientific and modern professional and technical training.
Fourth, improve the construction and allocation of school sports facilities. The sports venues of most colleges and universities need to be improved. Colleges and universities can expand their sports investment through selffinancing, social investment and fund-raising. For the maintenance and management funds of sports equipment, colleges and universities can promote sports teams to the society and market according to the principle of voluntariness under the normal academic arrangement of athletes, forming a diversified capital structure, which is an efficient way to solve the shortage of funds. However, the joint management of teams by social enterprises and the sports association is an ideal team management mode under the current situation.

Conclusion
The research of BP neural network in sports training has become a hot spot for scholars. In this paper, BP neural network is improved and a mathematical model of sports action classification is constructed, which further improves the accuracy of action classification. The application of visualization technology in the field of education will further promote the rapid development of education. Sports training can intuitively and dynamically observe athletes' movements and techniques with the help of visualization technology, which will greatly improve the training efficiency for coaches and athletes. Based on BP neural network and visualization technology, this paper designs a sports training management visualization system, including user management, training management, test management, plan making, scientific research information management and competition information management, and tests the visualization effect. Finally, it puts forward the problems existing in the management of physical training, including the backward educational concept of the school and the low level of informatization. Based on this, this paper puts forward the development strategy of physical training management. The school should strengthen the understanding of the overall goal and the training goal, be good at using the intelligent system to enhance the training efficiency of athletes, and the visual system of physical training management can help coaches and athletes improve their professional and technical level, strengthen the professional level and ability of coaches, especially the information literacy of coaches, be good at using intelligent information-based sports training equipment, improve the modernization level of sports training, improve the professional standards of athletes and expand the source of students, stimulate the enthusiasm of athletes for training and learning, improve the construction and allocation of school sports facilities, form a diversified capital structure, and improve the utilization rate of sports training equipment.
Funding The authors have not disclosed any funding.
Data availability Data will be made available on request.

Declarations
Conflict of interest The authors declare that they have no conflict of interests.
Ethical approval This article does not contain any studies with human participants performed by any of the authors.