College student physique test score prediction method based on CIWOA-BP neural network

CN118940887APending Publication Date: 2024-11-12GUILIN UNIV OF ELECTRONIC TECH
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410977290.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-22
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

The scoring system evaluation method of the traditional college student's physical fitness test scores cannot effectively reflect the students' physical health, and it lacks scientificity and objectivity when facing a large amount of data and complex factors.

Method used

Using a method based on CIWOA-BP neural network, the weight and threshold of the BP neural network are optimized through chaotic mapping Cubic map and adaptive weight optimization whale optimization algorithm (CIWOA) to construct a CIWOA-BP neural network model for predicting physical fitness test for college students score.

Benefits of technology

It improves prediction accuracy and generalization ability, can evaluate the physical health level of college students more accurately and objectively, and shows higher stability and robustness when facing complex data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118940887A_ABST
    Figure CN118940887A_ABST
Patent Text Reader

Abstract

The invention discloses a college student physique test score prediction method based on a CIWOA-BP neural network, and relates to the field of sports and computer science. The method comprises the following steps: collecting constitution tests of college students in a certain school year to form a data set, removing abnormal values of the data from the data set by adopting a 3 sigma principle, carrying out data standardization through a normalization method, and extracting part of data from the data set through layered sampling to divide a training set and a test set. The CIWOA algorithm is adopted to optimize the initial weight and threshold of the BP neural network model, the CIWOA algorithm utilizes Cubic map chaotic mapping to optimize the initial population of the WOA algorithm, and the adaptive weight is utilized to optimize the weight of the WOA algorithm shrinkage bounding mechanism. The training data are substituted into the CIWOA-BP neural network to establish a college student physique test prediction model, and then the test data are substituted into the trained model to output a simulation result. According to the invention, the college student physique test score can be accurately predicted, and a technical means is provided for establishing a scientific and objective college student physique test evaluation system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the fields of sports science and computer science, and in particular to a method for predicting physical fitness test scores of college students based on a CIWOA-BP neural network. Background Art

[0002] With the continuous development of higher education in my country, the number of college students in various universities is huge. Against the background of a huge number of college students, there are huge differences in the physical health and physical fitness of college students. The health and physical fitness of college students has gradually attracted widespread attention from all walks of life. It is not only related to personal development, but also determines the quality of national human resources. In 2014, the Ministry of Education revised the "National Student Physical Fitness Standards" to improve the physical health level of college students, promote sports work in major universities, and promote the all-round development of students. The measurement items of the "National Student Physical Fitness Standards" include height / weight, vital capacity, 50-meter run, standing long jump, sit-ups, 1000-meter (male) / 800-meter (female) run, pull-ups (male) / 1-minute sit-ups (female). When evaluating students' physical fitness, the traditional method is mainly to evaluate through a scoring system. Due to the different importance of each item in the scoring system and the different evaluation criteria of each school, this method cannot effectively reflect the physical health of students when faced with a large amount of data and complex factors. The method of analyzing college students' physical fitness tests through neural networks helps to obtain a more accurate, scientific and objective evaluation system for college students' physical fitness tests, and effectively solves the defects of the traditional scoring evaluation method.

[0003] BP (Back Propagation) neural network is a multi-layer feedforward neural network, and its core principle is to use the back propagation algorithm for supervised learning. Its advantage is that when the number of hidden layers and nodes is large enough, it can approximate any nonlinear mapping relationship and has good generalization ability. The topological structure of the BP neural network is mainly composed of input layer, hidden layer, and output layer, and each layer contains multiple neurons. In the forward propagation stage, the input data passes through each layer of neurons and finally obtains the predicted result in the output layer. If the predicted result deviates from the actual result, the neural network will enter the back propagation stage, determine the error by calculating the loss function (such as mean square error), and reversely transmit these errors to adjust the weights and thresholds between each layer of neurons. This process will continue to iterate until the prediction error of the neural network is reduced to the set range.

[0004] The Whale Optimization Algorithm (WOA) is a nature-inspired optimization algorithm proposed by Mirjalili et al. in 2016. It simulates the social behavior and hunting strategies of whales to find the optimal solution. In the Whale Optimization Algorithm, each whale represents a solution. Whales continuously explore and capture prey during the iteration of the algorithm, and constantly try to improve their solutions. The Whale Optimization Algorithm can effectively solve various complex optimization problems and has a fast convergence speed. The algorithm's optimization process is mainly divided into three stages: surrounding prey, capturing prey, and searching for prey.

[0005] Chaotic map is a mathematical model used to describe a system that is orderly on the surface but chaotic on the inside. Cubic map is a specific type of chaotic map that is based on a cubic equation and generates a series of values ​​through iterative calculations. In this model, the result of each iteration depends on the result of the previous iteration, and the choice of parameters directly affects the dynamic behavior of the system. When the parameters are properly adjusted, the cubic map can exhibit chaotic characteristics, that is, a small change in the initial conditions can lead to completely different evolutionary paths. This chaotic behavior is often used in science and engineering to help researchers analyze and understand uncertainty and unpredictability in complex systems. Chaotic mapping is introduced into the whale algorithm (WOA) to maintain the diversity of the whale population and improve the global search capability of WOA.

[0006] The purpose of adaptive inertia weight is to balance the global exploration (extensive search) and local exploration (deep search) of the algorithm. As the algorithm proceeds, adaptively adjusting the inertia weight can help the algorithm avoid falling into the local optimal solution while maintaining the ability to search for the global optimal solution. Introducing the adaptive inertia weight mechanism in the WOA algorithm can effectively balance the global search and local search capabilities of the WOA algorithm. This strategy helps avoid the algorithm from falling into the local optimal solution, while significantly improving the accuracy and efficiency of optimization. Summary of the invention

[0007] The present invention aims to solve the defects of the traditional scoring system for evaluating the physical fitness test scores of college students, and provides a method for predicting the physical fitness test scores of college students based on the CIWOA-BP neural network. The method optimizes the initial population and global search ability of WOA through chaotic mapping Cubicmap and adaptive weight I, and optimizes the weights and thresholds of the BP neural network through the CIWOA algorithm, thereby constructing a CIWOA-BP neural network model. The model proposed by the method has high prediction accuracy, good generalization ability and robustness, and can accurately evaluate the physical health level of college students.

[0008] The technical solution for achieving the purpose of the present invention is:

[0009] A method for predicting college students' physical fitness test scores based on CIWOA-BP neural network specifically includes the following steps:

[0010] Step 1: Collect the physical fitness test of college students in a certain academic year as a data set, which includes physical fitness test items and physical fitness test scores;

[0011] Step 2: Preprocess the collected physical fitness test data; the data preprocessing method is as follows:

[0012] Step 2-1: First, delete the data with missing values ​​in the data set, then use the 3σ principle to remove the outliers in the data set, and divide the physical fitness test scores in the data set into four levels: fail, pass, good, and excellent;

[0013] Step 2-2: Use the normalization method to standardize and scale the data. The normalized data all fall in the interval [0,1];

[0014] Step 2-3: Extract some data from the data set according to the proportion of physical fitness test scores through stratified sampling, and then use the holdout method to divide the training set and test set;

[0015] Step 3: Determine the topological structure, network parameters and optimal number of hidden layer nodes of the BP neural network;

[0016] Step 4: Introduce the Whale Optimization Algorithm (WOA), and optimize the WOA algorithm through the chaotic map Cubic map and the adaptive weight coefficient I to form the CIWOA algorithm;

[0017] Step 5: Determine the initial population size, maximum number of iterations and other parameters of the CIWOA algorithm, optimize the BP neural network with the CIWOA algorithm, construct the CIWOA-BP neural network, and bring the training data set into the CIWOA-BP neural network for training;

[0018] Step 6: Use the CIWOA-BP neural network model trained in step 5 for the test set, output the simulation prediction results, and compare the BP neural network model with the CIWOA-BP neural network model.

[0019] In step 1, the physical fitness test items mainly include height / weight, vital capacity, 50-meter run, standing long jump, sit-ups, 1000-meter (male) / 800-meter (female) run, pull-ups (male) / 1-minute sit-ups (female).

[0020] In step 3, the range of the number of hidden layer nodes in the topological structure of the BP neural network is given by the empirical formula Calculation, where m is the number of input layer nodes, n is the number of output layer nodes, and a is an integer between 1 and 10; the optimal number of hidden layer nodes is determined by the coefficient of determination (R 2 ) and root mean square error (MSE), R 2 The closer it is to 1, the closer MSE is to 0, indicating that the BP neural network model has a better fit and is more accurate. This is the optimal number of hidden layer nodes.

[0021] In step 4, the chaotic map Cubic map is used to optimize the initial population of the WOA algorithm to maintain the diversity of the whale population and improve the global search capability of the WOA algorithm; the adaptive weight coefficient I is used to optimize the weight of the shrinking and surrounding mechanism of the WOA algorithm to prevent the WOA algorithm from falling into the local optimal solution while maintaining the search capability for the global optimal solution.

[0022] In step 5, the purpose of CIWOA optimizing the BP neural network is to obtain better weights and thresholds, mainly by taking the test error of the BP neural network as the target fitness function F, finding the optimal target value through the CIWOA algorithm, that is, the optimal weight and threshold, and assigning the optimal target value to the BP neural network to construct the CIWOA-BP neural network.

[0023] The present invention discloses a method for predicting the physical fitness test scores of college students based on a CIWOA-BP neural network. The above technical scheme adopted has the following advantages: different from the WOA algorithm, the CIWOA algorithm introducing a chaotic map Cubic map and an adaptive weight coefficient I has a better global solution-finding ability; different from the BP neural network, the CIWOA-BP neural network introducing the CIWOA algorithm has higher prediction accuracy, generalization ability and robustness; different from the traditional scoring system for evaluating the physical fitness test scores of college students, when faced with a large amount of data and complex factors, the prediction method can accurately and objectively evaluate the physical fitness test scores of college students. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is a flow chart of the method of the present invention;

[0025] Figure 2 A distribution diagram of physical fitness test data of a certain university provided by an embodiment of the present invention;

[0026] Figure 3 A BP neural network topology diagram provided by an embodiment of the present invention;

[0027] Figure 4 A flow chart of a CIWOA-BP neural network provided by an embodiment of the present invention;

[0028] Figure 5 An evolution curve diagram of the CIWOA algorithm provided by an embodiment of the present invention;

[0029] Figure 6 A CIWOA-BP neural network training performance diagram provided by an embodiment of the present invention;

[0030] Figure 7 A comparison chart of the predicted values ​​and true values ​​of the BP and WOA-BP neural networks provided in an embodiment of the present invention;

[0031] Figure 8 A comparison chart of the prediction errors of the BP and WOA-BP neural networks provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The specific implementation of the present invention will be described in more detail below in conjunction with the schematic diagram, but the present invention is not limited thereto. The advantages and features of the present invention will become more apparent from the following description and claims. It should be noted that the drawings are all in a very simplified form and are not in precise proportions, and are only used to conveniently and clearly assist in explaining the purpose of the embodiments of the present invention.

[0033] Example:

[0034] like Figure 1 As shown, a method for predicting college students' physical fitness test scores based on CIWOA-BP neural network includes:

[0035] Step S1: The physical fitness test scores of college students in a certain academic year of a certain university were used as research samples. A total of 34,882 data samples were collected. Some of the collected data samples are shown in the following table:

[0036] Table 1 Data sample of college students' physical fitness test scores (partial)

[0037]

[0038] In Table 1, columns 3-10 are the physical fitness test items, and column 11 is the physical fitness test results.

[0039] Step S2: preprocessing the collected physical fitness test data; the data preprocessing method is as follows:

[0040] Step S2-1: Delete the data with missing values ​​in the data set, and use the 3σ principle to remove the data with outliers. The distribution of the processed physical fitness test data is as follows: Figure 2 As shown in the figure, there are 34707 data samples in total. The male and female data samples are divided into two different data sets, and the physical fitness test scores in the data set are divided into four levels, namely, failing (less than 60 points), passing (60-79 points), good (80-89 points), and excellent (greater than 90 points).

[0041] Step S2-2: Use the normalization method to standardize and scale the data. The normalized data all fall within the [0,1] interval. The normalization formula is as follows:

[0042]

[0043] In the formula, x is the normalized data; is the i-th value in the t-th column of the initial data; is the minimum value in the tth column of the initial data; is the maximum value in the tth column of the initial data.

[0044] Step S2-3: Through stratified sampling, 2000 groups of data are extracted from the male and female data sets according to the proportion of physical fitness test scores, and then the order of these 2000 groups of data is randomly arranged, and then 80% of the data are used as training sets and 20% of the data are used as test sets using the holdout method. As shown in Table 1, data No. 1 to 16 are training set data and data No. 17 to 20 are test set data.

[0045] Step S3: In this embodiment, the number of input layer nodes of the BP neural network is 8, the number of output layer nodes is 1, and the number of hidden layer nodes is calculated by an empirical formula, which is shown in the following formula:

[0046]

[0047] In the formula, m is the number of input layer nodes; n is the number of output layer nodes; a is an integer between 1 and 10; the number of hidden layer nodes can be calculated to be in the range of 4 to 13 through the empirical formula.

[0048] The coefficient of determination (R 2 ) and root mean square error (MSE) to determine the optimal number of hidden layer nodes, R 2 The closer it is to 1, the closer MSE is to 0, indicating that the BP neural network model has a better fit and is more accurate, which is the optimal number of hidden layer nodes. 2 The calculation formula of MSE is as follows:

[0049]

[0050] In the formula, y i Indicates the measured value; represents the fitted value; Represents the mean of all data; different numbers of hidden layer nodes are brought into the BP neural network for simulation testing, and the optimal number of hidden layer nodes is 13, and the corresponding root mean square error (MSE) is: 0.0072907, so the topological structure of the BP neural network is determined to be 8-13-1, such as Figure 3 As shown;

[0051] The network parameters of the BP neural network are set as follows: the number of trainings is 1000, the learning rate is 0.01, the minimum error of the training target is 0.0001, the display frequency is 25, the momentum factor is 0.01, the minimum performance gradient is 1e-6, and the maximum number of failures is 6.

[0052] Step S4: Introduce the whale optimization algorithm (WOA), the optimization process of which is mainly divided into three stages: surrounding prey, capturing prey and searching for prey;

[0053] In the stage of surrounding prey, the whale updates its position by surrounding the prey when hunting. The formula for surrounding the prey is:

[0054]

[0055] Where t is the current iteration number; T max is the maximum number of iterations; A and C are the representation coefficients; D is the bracketing step length; X * (t) is the whale position with the best current fitness; X(t) is the whale position in the current iteration; r 1 and r 2 is a random number in the interval [0,1]; a is a coefficient that changes linearly from 2 to 0;

[0056] In the prey capture phase, whales capture prey through spiral encirclement and contraction encirclement. The probability of these two methods is 50%. The formula for capturing prey is:

[0057]

[0058] Where D p is the distance between the whale and its prey; b is the spiral shape constant; l is a random number in the interval [-1,1];

[0059] In the prey search phase, when the coefficient A ≥ 1, the whale will randomly select a companion's position to approach, update the positions of other whales, increase the randomness of exploration, improve the global search capability, and find a better solution. The formula for searching prey is:

[0060]

[0061] Where, X rand is a random whale position;

[0062] The chaotic map Cubic map is introduced to optimize the initial population of the WOA algorithm. The Cubic map model is shown as follows:

[0063]

[0064] In the formula, C nis the nth chaotic value; β is the control parameter, which is usually 1;

[0065] The adaptive weight coefficient I is introduced to optimize the weight of the shrinking and surrounding mechanism of the WOA algorithm, as shown in the following formula:

[0066] ω=ω min +(ω max -ω min )*S*exp(-t / t max )

[0067] In the formula, ω min and ω max are the minimum and maximum values ​​of the adaptive weight respectively; S is the adjustment coefficient; t is the current iteration number; t max is the maximum number of iterations.

[0068] Step S5: Set the parameters of the CIWOA algorithm as follows: the initial population size is 50, the maximum evolutionary generation is 300, the lower limit of the independent variable is 0.1, and the upper limit of the independent variable is 2;

[0069] The test error of the BP neural network is used as the target fitness function F to bring it into the CIWOA algorithm to find the optimal target value (weight, threshold), and the optimal target value is assigned to the BP neural network to construct the CIWOA-BP neural network. The specific process is as follows: Figure 4 As shown;

[0070] The training set divided in step S2 is brought into the CIWOA-BP neural network for training, and the CIWOA evolution curve is obtained as follows: Figure 5 As shown in the figure, when the number of iterations of the male model reaches 57 and the number of iterations of the female model reaches 93, the mean square errors are 2.05768 and 1.68767 respectively, and the mean square error no longer changes with the number of iterations, indicating that the maximum number of iterations has been reached, and this is the optimal whale individual;

[0071] The prediction performance of CIWOA-BP neural network is as follows Figure 6 As shown in the figure, we can see that after optimization by the whale optimization algorithm, the training regression coefficient R of the CIWOA-BP neural network physical fitness test prediction model for boys and girls are 0.99026 and 0.98775 respectively. The closer the coefficient is to 1, the better the model fit and the more accurate it is. It can be seen that the fitting effect of the CIWOA-BP neural network college students' physical fitness test prediction model is good.

[0072] Step S6: Apply the BP neural network and CIWOA-BP neural network prediction models to the test data set of college students' physical fitness test scores, and compare the predicted values ​​with the true values, such as Figure 7As shown in the figure, it can be seen that the predicted values ​​of the CIWOA-BP neural network are well consistent with the true values ​​in most test samples, and it is better than the BP neural network in the prediction of test samples with large fluctuations;

[0073] The prediction comparison results for the test sample with the smallest index value are shown in the following table:

[0074] Table 2 Prediction of the minimum index value of the test sample

[0075]

[0076] From Table 2, we can see that the accuracy of the BP neural network prediction model for boys and girls is 88.6% and 65.2% respectively, while the accuracy of the CIWOA-BP neural network prediction model for boys and girls is 97.1% and 83.2% respectively, which shows that when facing data with large fluctuations, the CIWOA-BP neural network has a higher degree of fit and better prediction effect than the BP neural network;

[0077] The prediction errors of BP and CIWOA-BP neural network models are compared. The comparison chart of BP and CIWOA-BP neural network prediction errors is as follows: Figure 8 As shown in the figure, it can be seen that the prediction error of the BP neural network is relatively unstable, and the prediction error value is large in some test samples. The prediction error value of the CIWOA-BP neural network is generally concentrated in the zero scale line area, and the volatility is smaller than that of the BP neural network;

[0078] The maximum prediction errors of BP and CIWOA-BP neural network models are shown in the following table:

[0079] Table 3 Maximum prediction error of the model

[0080]

[0081] From Table 3, we can see that for the physical test error of boys, the maximum prediction errors of BP neural network and CIWOA-BP neural network are 33.7% and 28.7% respectively; for the physical test error of girls, the maximum prediction errors of BP neural network and CIWOA-BP neural network are 34.8% and 16.8% respectively. It can be seen that the prediction error of CIWOA-BP neural network is smaller than that of BP neural network, and it has strong robustness and generalization ability, and the prediction results are closer to the true value, and the prediction effect is better.

[0082] The implementation process of the above content is completed in MATLAB, and the process is also applicable to other programming software, such as Python.

[0083] In summary, the present invention collects the physical fitness test scores of college students in a certain academic year of a certain university as research samples, pre-processes the collected data through methods such as the 3σ principle, data normalization, stratified sampling and holdout method, and adopts chaotic map Cubic map, adaptive weight coefficient I and WOA algorithm to form a CIWOA algorithm, and then uses the test error of the BP neural network as the target fitness function F, and uses the CIWOA algorithm to find the optimal weight threshold and assign it to the BP neural network to construct a CIWOA-BP neural network model, and then brings the training set into the CIWOA-BP neural network to construct a prediction model for the physical fitness test scores of college students based on the CIWOA-BP neural network, and finally brings the test set into the test set to output the simulation results; from the performance of the test set, the prediction accuracy and generalization ability of the prediction model for the physical fitness test scores of college students based on the CIWOA-BP neural network are better, and it shows stronger stability when facing noise data and outliers. The model of this method can accurately predict the physical test scores of college students.

[0084] The above is only a preferred embodiment of the present invention and does not limit the present invention in any way. Any technician in the relevant technical field, without departing from the scope of the technical solution of the present invention, makes any form of equivalent replacement or modification to the technical solution and technical content disclosed in the present invention, which does not depart from the content of the technical solution of the present invention and still falls within the protection scope of the present invention.

Claims

1. A method for predicting college students' physical fitness test scores based on CIWOA-BP neural network, characterized in that: The following steps are involved: Step 1: Collect the physical fitness test of college students in a certain academic year as a data set; Step 2: Preprocess the collected physical fitness test data; the data preprocessing method is as follows: Step 2-1: First delete the data with missing values ​​in the data set, and then remove the outliers in the data set; Step 2-2: Use normalization method to scale the data; Step 2-3: Use the holdout method to divide the data set into training set and test set; Step 3: Determine the topological structure, network parameters and optimal number of hidden layer nodes of the BP neural network; Step 4: Introduce the Whale Optimization Algorithm (WOA), and construct the CIWOA algorithm through the chaotic map Cubic map and the adaptive weight coefficient I; Step 5: Determine the parameters such as the initial population size of the CIWOA algorithm, combine the CIWOA algorithm with the BP neural network, construct the CIWOA-BP neural network, and bring the training data set into the CIWOA-BP neural network for training; Step 6: Use the CIWOA-BP neural network model trained in step 5 for the test set, output the simulation prediction results, and then compare the BP neural network model with the CIWOA-BP neural network model.

2. According to claim 1, a method for predicting college students' physical fitness test scores based on CIWOA-BP neural network is characterized by: In step 1, the collected data set includes physical fitness test items and physical fitness test scores, wherein the physical fitness test items mainly include height / weight, vital capacity, 50-meter run, standing long jump, sit-ups, 1000-meter (male) / 800-meter (female) run, pull-ups (male) / 1-minute sit-ups (female).

3. The method for predicting college students' physical fitness test scores based on CIWOA-BP neural network according to claim 1 is characterized by: In step 3, the empirical formula Calculate the range of the number of hidden layer nodes of the BP neural network, where m is the number of input layer nodes, n is the number of output layer nodes, and a is an integer between 1 and 10.

4. According to claim 3, a method for predicting college students' physical fitness test scores based on CIWOA-BP neural network is characterized in that: After the range of hidden layer nodes is determined, the determination coefficient (R 2 ) and root mean square error (MSE) to determine the optimal number of hidden layer nodes, R 2 The closer it is to 1, the closer MSE is to 0, indicating that the BP neural network model has a better fit and is more accurate. This is the optimal number of hidden layer nodes.

5. The method for predicting college students' physical fitness test scores based on CIWOA-BP neural network according to claim 1 is characterized by: In step 4, the chaotic map Cubic map is used to optimize the initial population of the WOA algorithm, and the adaptive weight coefficient I is used to optimize the weight of the shrinking and surrounding mechanism of the WOA.

6. The method for predicting college students' physical fitness test scores based on CIWOA-BP neural network according to claim 1 is characterized by: In step 5, the purpose of CIWOA optimizing the BP neural network is to obtain better weights and thresholds, mainly by taking the test error of the BP neural network as the target fitness function F, solving the optimal target value through the CIWOA algorithm, that is, the optimal weight and threshold, and assigning the optimal target value to the BP neural network to construct the CIWOA-BP neural network.

Citation Information

Cited By

  • Photovoltaic curtain wall system power generation efficiency prediction method and system

    CN120357441A