Intelligent Monitoring and Abnormality Recognition Method, System and Storage Medium for Strength Training

Abstract

The present invention belongs to the technical field of training data processing, and specifically provides a method, system and storage medium for intelligent monitoring and abnormal recognition of strength training. The method mainly includes: obtaining the target and real-time training parameters, calculating the training density and the intermittent and intensity coupling factor, generating a dynamic compensation coefficient and an initial intensity correction gradient through a random forest and constraining them, decomposing the deviation rate into a trend term and a mutation term by using an attention network, allocating weights to enhance the trend term, and outputting an efficacy index and a risk level according to stability and peak values. The present invention can effectively solve the problem that in the prior art, fixed-parameter modes are mostly used in strength training, mainly executed according to pre-designed plans, and it is difficult to accurately evaluate the training efficacy and identify abnormal risks. It can accurately evaluate the training efficacy, effectively identify abnormal risks, significantly improve the scientificity of training, optimize the training effect, and ensure the safety of the training process.

Description

Technical Field

[0001] The present invention belongs to the technical field of training data processing, and particularly relates to a method, system and storage medium for intelligent monitoring and abnormal recognition of strength training. Background Art

[0002] Strength training enhances muscle strength and endurance, increases the basal metabolic rate, promotes bone health, improves body posture, and enhances sports ability and quality of life. Monitoring strength training can scientifically evaluate the training effect, accurately adjust the training plan, effectively prevent sports injuries, and ensure the safety and efficiency of training.

[0003] Currently, most strength training adopts a fixed parameter mode, mainly executed according to a pre-designed plan, and it is difficult to capture dynamic information such as the change of the inter-group interval and the fluctuation of the training intensity in real time, resulting in the difficulty of dynamically adjusting training parameters according to the actual situation. As a result, it is impossible to conduct in-depth analysis based on accurate data, and it is difficult to accurately evaluate the training efficiency and identify abnormal risks. Summary of the Invention

[0004] The present disclosure effectively solves the problem that in the prior art, most strength training adopts a fixed parameter mode, mainly executed according to a pre-designed plan, and it is difficult to accurately evaluate the training efficiency and identify abnormal risks, by providing a method, system and storage medium for intelligent monitoring and abnormal recognition of strength training. It can accurately evaluate the training efficiency, effectively identify abnormal risks, significantly improve the scientific nature of training, optimize the training effect, and ensure the safety of the training process.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] In a first aspect, the present disclosure provides a method for intelligent monitoring and abnormal recognition of strength training, including: obtaining target training parameters and real-time training parameters; calculating training density and the coupling factor of interval and intensity according to the target training parameters, processing the training density and the coupling factor of interval and intensity by using a random forest model, outputting a dynamic compensation coefficient and an intensity correction gradient, and performing a non-linear constraint on the intensity correction gradient; processing the dynamic compensation coefficient, real-time training parameters and intensity correction gradient by using an attention gating network, including: the real-time training parameters include the actual inter-group interval deviation rate, performing time series decomposition on the actual inter-group interval deviation rate by using the STL decomposition method to generate a trend term and a mutation term; allocating the weights of the attention heads by using the dynamic compensation coefficient and the intensity correction gradient, and enhancing the features of the trend term by using a multi-head attention mechanism to generate a trend enhancement term; outputting a training efficiency index and an abnormal risk level according to the stability of the trend enhancement term and the peak value of the mutation term.

[0007] Further, the target training parameters include the training volume, training intensity, target time for rest between sets, and target time for rest within a set; calculating the training density and the rest-intensity coupling factor based on the target training parameters includes: dividing the training volume by the product of the target time for rest between sets and the target time for rest within a set to obtain the training density; multiplying the difference between the target time for rest between sets and the target time for rest within a set by the training intensity to obtain the rest-intensity coupling factor.

[0008] Further, a random forest model is used to process the training density and the rest-intensity coupling factor, and a dynamic compensation coefficient and an intensity correction gradient are output, including: normalizing the training density and the rest-intensity coupling factor as the dynamic feature weights of each decision tree; based on the dynamic feature weights, using the CART algorithm to perform node splitting to generate the dynamic compensation coefficient and the intensity correction gradient.

[0009] Further, the real-time training parameters also include the load weight, and the load volatility is calculated based on the load weight; performing non-linear constraint on the intensity correction gradient includes: setting the base range, the first threshold, and the minimum upper limit value of the intensity correction gradient, and when the load volatility exceeds the first threshold, reducing the upper limit value of the base range to the minimum upper limit value; performing proportional scaling on the intensity correction gradient and the actual rest-between-sets deviation rate and then performing non-linear compression; using the gradient clipping algorithm to perform boundary constraint on the compressed intensity correction gradient based on the base range.

[0010] Further, dividing the difference between the load weight and the training intensity by the training intensity to obtain the load volatility.

[0011] Further, the real-time training parameters also include the actual rest-within-a-set deviation rate; using the STL decomposition method to perform time series decomposition on the actual rest-between-sets deviation rate to generate a trend term and a mutation term, including: filtering outliers and filling missing values in the actual rest-between-sets deviation rate to generate time series data; determining the decomposition window length according to the length of the time series data, and performing smoothing processing on the data within each window to output the trend term; taking the difference between the time series data and the trend term as the mutation term, and adjusting the amplitude of the mutation term based on the dynamic compensation coefficient.

[0012] Further, the weights of the attention heads are allocated through the dynamic compensation coefficient and the intensity correction gradient, and the multi-head attention mechanism is used to enhance the features of the trend term to generate a trend-enhanced term, including: performing weighted summation on the dynamic compensation coefficient and the intensity correction gradient to generate the allocation weights of each attention head in the multi-head attention mechanism; using the multi-head attention mechanism to enhance the features of the trend term, and performing normalized weighted fusion on the output results of multiple attention heads to generate the trend-enhanced term.

[0013] Further, the training efficiency index and the abnormal risk level are trained according to the stability of the trend enhancement term and the peak output of the mutation term, including: dividing windows for the trend enhancement term, calculating the variance of all trend enhancement terms included in each window as the stability index; extracting the maximum absolute value of the mutation term in the time series as the peak value; assigning weights to the stability index and the peak value, linearly mapping the weighted sum result of the stability index and the peak value amplitude to a specified interval to obtain the training efficiency index; and outputting the abnormal risk level according to the peak value and the dynamic compensation coefficient.

[0014] In a second aspect, the present disclosure provides a strength training intelligent monitoring and abnormal identification system, which includes: a parameter acquisition module, a factor calculation and model derivation module, and a real-time data processing and evaluation output module.

[0015] The parameter acquisition module is used to acquire target training parameters and real-time training parameters.

[0016] The factor calculation and model derivation module is used to calculate the training density and the intermittent and intensity coupling factor according to the target training parameters, process the training density and the intermittent and intensity coupling factor by using a random forest model, output the dynamic compensation coefficient and the initial intensity correction gradient, and perform a non-linear constraint on the intensity correction gradient;

[0017] The real-time data processing and evaluation output module is used to process the dynamic compensation coefficient, the real-time training parameters, and the intensity correction gradient by using an attention gating network, including: the real-time training parameters include the actual inter-group intermittent deviation rate, perform time series decomposition on the actual inter-group intermittent deviation rate by using the STL decomposition method to generate a trend term and a mutation term; allocate the weights of the attention heads through the dynamic compensation coefficient and the intensity correction gradient, use the multi-head attention mechanism to enhance the features of the trend term to generate a trend enhancement term; output the training efficiency index and the abnormal risk level according to the stability of the trend enhancement term and the peak value of the mutation term.

[0018] In a third aspect, the present disclosure provides a strength training intelligent monitoring and abnormal identification device, which includes a memory and a processor; the memory is used to store a computer program; the processor is used to implement the steps of the strength training intelligent monitoring and abnormal identification method as described in the first aspect when executing the computer program.

[0019] In a fourth aspect, the present disclosure provides a storage medium, in which computer program instructions are stored, and when the computer program instructions are read and run by a processor, the steps of the strength training intelligent monitoring and abnormal identification method as described in the first aspect are executed.

[0020] Advantages of the present invention:

[0021] The present invention solves the problem in the prior art that strength training mostly adopts a fixed-parameter mode and is mainly carried out according to a pre-designed plan, making it difficult to accurately evaluate training effectiveness and identify abnormal risks. By obtaining the target and real-time training parameters, calculating the training density and the coupling factor of the interval and intensity, generating and constraining the dynamic compensation coefficient and the initial intensity correction gradient through a random forest, decomposing the deviation rate into a trend term and a mutation term using an attention network, allocating weights to enhance the trend term, and outputting the effectiveness index and risk level according to stability and peak value, the present invention can accurately evaluate training effectiveness, effectively identify abnormal risks, significantly improve the scientific nature of training, optimize the training effect, and ensure the safety of the training process.

[0022] Other features and advantages of the present invention will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained by the structure pointed out in the specification and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0024] Figure 1 The flowchart of the intelligent monitoring and abnormal identification method for strength training according to the present invention is shown;

[0025] Figure 2 The block diagram of the intelligent monitoring and abnormal identification system for strength training according to the present invention is shown. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] In order to solve the problem in the prior art that strength training mostly adopts a fixed-parameter mode and is mainly carried out according to a pre-designed plan, making it difficult to accurately evaluate training effectiveness and identify abnormal risks, the present disclosure obtains the target and real-time training parameters, calculates the training density and the coupling factor of the interval and intensity, generates and constrains the dynamic compensation coefficient and the initial intensity correction gradient through a random forest, decomposes the deviation rate into a trend term and a mutation term using an attention network, allocates weights to enhance the trend term, and outputs the effectiveness index and risk level according to stability and peak value, so as to accurately evaluate training effectiveness, effectively identify abnormal risks, significantly improve the scientific nature of training, optimize the training effect, and ensure the safety of the training process.

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0028] In some embodiments, as Figure 1 shown, the present disclosure provides a method for intelligent monitoring and abnormal recognition of strength training, including:

[0029] S100. Obtain target training parameters and real-time training parameters, where the real-time training parameters include the actual inter-set rest deviation rate.

[0030] S200. Calculate the training density and the rest-intensity coupling factor according to the target training parameters, process the training density and the rest-intensity coupling factor using a random forest model, output a dynamic compensation coefficient and an intensity correction gradient, and perform a non-linear constraint on the intensity correction gradient.

[0031] S300. Process the dynamic compensation coefficient, the real-time training parameters, and the intensity correction gradient using an attention gating network, including:

[0032] S310. Perform a time series decomposition on the actual inter-set rest deviation rate using the STL decomposition method to generate a trend term and a mutation term.

[0033] S320. Allocate the weights of the attention heads through the dynamic compensation coefficient and the intensity correction gradient, and use the multi-head attention mechanism to enhance the features of the trend term to generate a trend-enhanced term.

[0034] S330. Output a training efficiency index and an abnormal risk level according to the stability of the trend-enhanced term and the peak value of the mutation term.

[0035] In some embodiments, the target training parameters in S100 include the training volume, the training intensity, the target inter-set rest time, and the target intra-set rest time.

[0036] ; where V represents the training volume, S represents the number of sets, R represents the number of repetitions per set, and W represents the weight of a single load. Exemplarily, if the user completes 3 sets of squats, 10 repetitions per set, and a load of 100 kg each time, then the training volume is: .

[0037] The training intensity represents the percentage of the load weight of a single movement to the user's maximum strength. Exemplarily, if the user's 1RM for squats is 150 kg and 120 kg is used in this training, then the training intensity is .

[0038] The target inter-group interval time is the preset rest duration between two groups of training, which is used to control the training density and recovery rhythm.

[0039] The target intra-group interval time is the preset interval time between two consecutive actions within the same group, which is used to maintain the action rhythm.

[0040] In some embodiments, calculating the training density and the interval-intensity coupling factor according to the target training parameters in S200 includes:

[0041] Dividing the training volume by the product of the target inter-group interval time and the target intra-group interval time to obtain the training density.

[0042] Multiplying the difference between the target inter-group interval time and the target intra-group interval time by the training intensity to obtain the interval-intensity coupling factor.

[0043] ; where D represents the training density, V represents the training volume, represents the target inter-group interval time, represents the target intra-group interval time.

[0044] ; where C represents the interval-intensity coupling factor and I represents the training intensity.

[0045] Exemplarily, when the user performs squat training, the training volume is times, the target inter-group interval time is 120 seconds, and the target intra-group interval time is 60 seconds, then the training density is: .

[0046] The training density reflects the completion of the training volume per unit time and can assist in adjusting the training rhythm.

[0047] The inter-group rest time is relatively long, aiming to allow the body to recover to a greater extent from the fatigue of a group of training so as to be able to perform the next group of training in a better state. The intra-group interval time is relatively short, mainly for briefly adjusting breathing and preparing for the next action, and the body's recovery degree is limited.

[0048] By subtracting the target intra-group interval time from the target inter-group interval time, the difference in the recovery degree can be reflected. The larger the difference, the more sufficient time there is between groups to recover physical strength and function and prepare for the next group of high-intensity training.

[0049] The training intensity directly determines the load borne by the muscles, and the interval-intensity coupling factor can predict the muscle fatigue rate to a certain extent.

[0050] In some embodiments, processing the training density and the intermittent and intensity coupling factor using a random forest model and outputting a dynamic compensation coefficient and an intensity correction gradient in S200 includes:

[0051] S210. Normalize the training density and the intermittent and intensity coupling factor as the dynamic feature weights of each decision tree.

[0052] S220. Based on the dynamic feature weights, use the CART algorithm for node splitting to generate a dynamic compensation coefficient and an intensity correction gradient.

[0053] The Min-Max normalization method can be used to process the training density and the intermittent and intensity coupling factor, and the normalized training density and intermittent and intensity coupling factor are used as the dynamic feature weights of each decision tree.

[0054] Select an optimal feature and an optimal splitting point on this feature from the training density and the intermittent and intensity coupling factor according to the dynamic feature weights. The optimal splitting point can be found with the goal of minimizing the mean squared error.

[0055] In each decision tree, a node represents a test on the training density or the intermittent and intensity coupling factor, a branch represents the result of the test, and a leaf node represents a dynamic compensation coefficient and an intensity correction gradient.

[0056] When constructing a decision tree, each node is evaluated.

[0057] When evaluating a node, calculate the mean squared error between the generated result and the actual target value when different values of the training density and the intermittent and intensity coupling factor are used as splitting points respectively. For example, if the mean squared error is smaller when the training density is 0.6 as the splitting point, then 0.6 is the optimal splitting point of this node based on the training density feature. The data set is divided accordingly to form the left and right child nodes of the decision tree.

[0058] For the generated left and right child nodes, continuously repeat the operations of feature selection and node splitting until the pre-set stop condition is met.

[0059] The stop condition can be that the depth of the tree reaches a preset value, such as setting the maximum depth to 5 layers; or it can be that the number of samples in a node is less than a certain threshold, for example, stipulating to stop splitting when the number of node samples is less than 10.

[0060] Through continuous recursive construction, a complete decision tree structure is formed. Many decision trees together form a random forest. Finally, the output results of multiple decision trees are synthesized, such as taking the average value, so as to generate a dynamic compensation coefficient and an intensity correction gradient.

[0061] The intensity correction gradient is used for subsequent attention mechanisms and anomaly detection.

[0062] The intensity correction gradient is used to dynamically adjust the training intensity according to real-time training data (such as load fluctuations, inter-group interval deviations, etc.), helping the trainer to automatically reduce the intensity adjustment range when the load fluctuates greatly to ensure training safety; in a stable state, a larger range of intensity optimization is allowed to improve the effect.

[0063] In some embodiments, the real-time training parameter in S100 further includes the load weight, and the load volatility can be calculated according to the load weight.

[0064] The non-linear constraint on the intensity correction gradient in S200 includes:

[0065] S230. Set the basic range, the first threshold, and the lowest upper limit value of the intensity correction gradient. When the load volatility exceeds the first threshold, reduce the upper limit value of the basic range to the lowest upper limit value.

[0066] S240. Perform proportional scaling on the intensity correction gradient and the actual inter-group interval deviation rate and then perform non-linear compression.

[0067] S250. Adopt a gradient clipping algorithm to perform boundary constraint on the compressed intensity correction gradient based on the basic range.

[0068] The basic range can be , the first threshold can be 15%, and the lowest upper limit value can be 8%.

[0069] When the load volatility , the upper limit value of the basic range changes from to , which can effectively avoid excessive intensification under high fluctuations, thereby reducing the risk of injury.

[0070] The proportional scaling formula can be: ; where represents the scaled intensity correction gradient, represents the intensity correction gradient before scaling, represents the actual inter-group interval deviation rate.

[0071] The non-linear compression formula can be: ; where represents the intensity correction gradient after non-linear compression by the tanh function, represents the upper limit value of the basic range.

[0072] For example, if the intensity correction gradient is and the actual inter-group interval deviation rate is 20%, then the scaled intensity correction gradient is: .

[0073] The gradient clipping algorithm is adopted to force the gradient after compression to be within the dynamically adjusted base range. The rules are as follows:

[0074] If the strength-corrected gradient after compression is greater than the upper limit of the base range, the strength-corrected gradient after compression takes the upper limit of the base range; if the strength-corrected gradient after compression is less than the lower limit of the base range, the strength-corrected gradient after compression takes the lower limit of the base range.

[0075] Gradient clipping can effectively ensure that the corrected gradient is strictly within the safe range.

[0076] In some embodiments, the difference between the load weight and the training intensity is divided by the training intensity to obtain the load volatility. The reference formula is: ; where F represents the load volatility, W represents the actual load weight, and I represents the training intensity.

[0077] In some embodiments, the real-time training parameters in S100 further include the actual intra-group interval deviation rate.

[0078] The actual intra-group interval deviation rate represents the deviation percentage of the actual interval time between two adjacent actions in the same group of training from the preset intra-group interval target time, and is calculated after obtaining the actual intra-group interval time.

[0079] The calculation formula is: , where represents the actual inter-group interval deviation rate, represents the actual inter-group interval time, the actual intra-group interval time represents the actual interval time between two consecutive actions within the same group, represents the inter-group interval target time.

[0080] The training status of the trainer changes continuously over time, and the actual inter-group interval deviation rate reflects the rhythm change situation during the training process.

[0081] In some embodiments, the attention gating network in S300 can capture the trends and mutation information during the training process by deeply analyzing this deviation rate, combining the dynamic compensation coefficient and the strength-corrected gradient, so as to provide more targeted training suggestions for coaches and trainers.

[0082] For example, when the trend term shows that the training intensity gradually and steadily increases, and the mutation term shows an abnormal peak, it may indicate that the trainer is at risk of fatigue or injury. At this time, the system can output the abnormal risk level in a timely manner and adjust the training plan.

[0083] The input of the attention gating network is the dynamic compensation coefficient, real-time training parameters, and strength-corrected gradient, and the output is the training efficiency index and the abnormal risk level.

[0084] The training efficiency index can comprehensively reflect the effect and efficiency of training, providing a quantitative indicator for evaluating training quality. The abnormal risk level can classify the possible risks according to the abnormal situations during the training process, helping coaches and training personnel take timely countermeasures.

[0085] The attention gating network consists of the following core modules: the time series decomposition module, the weight allocation module, the multi-head attention layer, and the evaluation module.

[0086] The time series decomposition module is used to receive the actual inter-group interval deviation rate in the real-time training parameters and output the trend term and the mutation term by using the STL decomposition method.

[0087] The weight allocation module is used to receive the dynamic compensation coefficient and the intensity correction gradient and calculate the weights of the multi-head attention heads.

[0088] The multi-head attention layer is used to enhance the features of the trend term and generate the trend-enhanced term.

[0089] The evaluation module is used to output the training efficiency index and the abnormal risk level according to the stability of the trend-enhanced term and the peak value of the mutation term.

[0090] The connection relationship of the core modules is as follows:

[0091] The actual inter-group interval deviation rate is input to the time series decomposition module, the decomposed trend term is input to the multi-head attention layer, and the mutation term is input to the evaluation module.

[0092] The dynamic compensation coefficient and the intensity correction gradient are input to the weight allocation module, and after generating the attention head weights, they are passed to the multi-head attention layer.

[0093] The trend-enhanced term output by the multi-head attention layer and the mutation term are jointly input to the evaluation module to generate the final result.

[0094] In some embodiments, in S310, the STL decomposition method is used to perform time series decomposition on the actual inter-group interval deviation rate to generate the trend term and the mutation term, including:

[0095] S311. Filter the outliers and fill the missing values of the actual inter-group interval deviation rate to generate time series data.

[0096] The data of the actual inter-group interval deviation rate may have outliers, such as extreme values and missing values caused by measurement errors. The outliers can be filtered by statistical methods, such as outlier detection based on the standard deviation, and the missing values can be filled by the interpolation method to generate continuous time series data , , where represents the T-th actual inter-group interval deviation rate, and T represents the number of actual inter-group interval deviation rates.

[0097] S312. Determine the window length for decomposition according to the length of the time series data, perform smoothing processing on the data within each window, and output the trend term.

[0098] Determine an appropriate window length according to the length of the time series data. For example, the LOWESS algorithm of STL decomposition can be used to set the periodic window length L, L = , after determining the window length, perform smoothing processing on the data within the window, and set the smoothing parameter , through the calculation process of iterating 3 times, finally output the trend term .

[0099] S313. Take the difference between the time series data and the trend term as the mutation term, and adjust the amplitude of the mutation term based on the dynamic compensation coefficient.

[0100] Mutation term , through the dynamic compensation coefficient Adjust the amplitude, and the mutation term after adjusting the amplitude is .

[0101] The dynamic compensation coefficient will amplify or reduce the value of the mutation term. For example, when the compensation coefficient is high, abnormal fluctuations will be marked more significantly.

[0102] In some embodiments, in S320, the weights of the attention heads are assigned through the dynamic compensation coefficient and the intensity correction gradient, and the multi-head attention mechanism is used to enhance the features of the trend term to generate a trend-enhanced term, including:

[0103] S321. Perform weighted summation of the dynamic compensation coefficient and the intensity correction gradient to generate the assignment weights for each attention head in the multi-head attention mechanism.

[0104] The assignment weights of the dynamic compensation coefficient and the intensity correction gradient can be obtained by grid search and cross-validation, comparing the model performance of different weight combinations on the historical training dataset.

[0105] Exemplarily, if the weight coefficients of the dynamic compensation coefficient and the intensity correction gradient are 0.6 and 0.4 respectively, then the assignment weight of each attention head is obtained by weighted summation of the dynamic compensation coefficient and the intensity correction gradient followed by Softmax normalization. Refer to the formula: .

[0106] S322. Use the multi-head attention mechanism to enhance the features of the trend term, and perform normalized weighted fusion on the output results of multiple attention heads to generate a trend-enhanced term.

[0107] Each attention head maps the trend term into query matrix, key matrix, and value matrix through linear transformation. The query matrix is multiplied by the transpose of the key matrix, and after scaling, the attention weights are generated through the Softmax function. The value matrix is weighted and summed using the attention weights to output the single-head attention result.

[0108] Using the allocation weights of each attention head, the outputs of multiple attention heads are weighted and summed according to the weights to finally generate the trend enhancement term.

[0109] The dynamic compensation coefficient is combined with the intensity correction gradient to determine the degree of focus of the model on the trend term. For example, when the compensation coefficient is high, the model pays more attention to the long-term stable trend and weakens the short-term noise.

[0110] In some embodiments, in S330, the training efficiency index and the abnormal risk level are output according to the stability of the trend enhancement term and the peak value of the mutation term, including:

[0111] S331. Perform window partitioning on the trend enhancement term, and calculate the variance of all trend enhancement terms included in each window as the stability index.

[0112] S332. Extract the maximum absolute value of the mutation term in the time series as the peak value.

[0113] S333. Assign weights to the stability index and the peak value, linearly map the weighted sum result of the stability index and the peak value amplitude to a specified interval to obtain the training efficiency index; output the abnormal risk level according to the peak value and the dynamic compensation coefficient.

[0114] Calculate the variance of all trend enhancement terms included in the window according to the variance calculation formula, and normalize the variance in each window to obtain the stability index. The formula can be: S = , where S represents the stability index, represents the variance of all trend enhancement terms included in the current window, represents the maximum value among all window variances.

[0115] If the mutation term is sequence M, the peak value is the maximum value in sequence M.

[0116] The stability index and the peak value can be weighted and summed with 50% weight each, and then linearly mapped to a specified interval, such as the 0-100 interval, to obtain the training efficiency index. The formula can be:

[0117]

[0118] Among them, E represents the training efficiency index, S represents the stability index, P represents the peak value, represents the preset peak upper limit.

[0119] Exemplarily, the risk levels may include high risk, medium risk, and low risk.

[0120] When the peak value and the dynamic compensation coefficient meet the conditions, it is determined as high risk.

[0121] When or meet the conditions, it is determined as medium risk.

[0122] When the conditions for high risk and medium risk are not met, it is determined as low risk.

[0123] When training the attention gating network, the dynamic compensation coefficient, real-time training parameters, and intensity correction gradient are used as inputs, and the trend enhancement term, training efficiency index, and abnormal risk level are used as outputs.

[0124] The mean squared error can be used for the regression task of the training efficiency index, and the cross-entropy loss can be used for the classification task of the abnormal risk level, etc. The Adam optimization algorithm is used to minimize the loss function.

[0125] During the training process, the parameters of the attention gating network are continuously adjusted to make the output of the network as close to the true value as possible. The gradient of the loss function with respect to the network parameters is calculated through the backpropagation algorithm, and the parameters are updated according to the optimization algorithm.

[0126] Through multiple iterations, the parameters of the network are gradually adjusted until the loss function converges to a smaller value or reaches the preset number of training epochs.

[0127] The training efficiency index can reflect the overall effect and efficiency of the current training. If the index is low, it indicates that the training effect may not be ideal, and it may be necessary to consider adjusting the training plan, such as changing parameters such as training intensity, rest interval, or training volume, to improve the training efficiency.

[0128] For example, if it is found that the long rest interval between groups leads to low training density, which in turn affects the training efficiency index, the rest interval between groups can be appropriately shortened.

[0129] The risk level can directly indicate the potential risks existing in the training process. If it is in a high-risk or medium-risk state, it is necessary to take timely measures to adjust the training to avoid injury or overtraining.

[0130] For example, when it is determined as high risk, it may be necessary to immediately stop the current training and conduct a comprehensive evaluation and adjustment of the training plan; if it is medium risk, the training intensity can be appropriately reduced or the rest interval can be increased to reduce the risk.

[0131] Different trainers have different physical conditions and training goals. By using the training efficiency index and risk level, a personalized training plan can be formulated for each trainer. According to the training efficiency index, understand the adaptation of each trainer to different training parameters, so as to optimize the training plan and make it more in line with personal characteristics.

[0132] For example, for a trainer with a relatively high training efficiency index but also a relatively high risk level, the risk can be reduced by adjusting the training parameters while maintaining the training effect; for a trainer with a relatively low training efficiency index, the training method can be adjusted specifically to improve the training effect.

[0133] Continuously recording the training efficiency index and risk level can track and evaluate the long-term training effect and physical state of the trainer.

[0134] Observe the change trend of the training efficiency index to understand whether the training plan is effective in the long term and whether the trainer's body adapts to this training method. At the same time, according to the change of the risk level, timely discover possible problems in the training process, such as overtraining, accumulation of physical fatigue, etc., so as to take preventive measures in advance to ensure the physical health of the trainer and the continuity of training.

[0135] In some embodiments, as Figure 2 shown, the present disclosure provides a strength training intelligent monitoring and anomaly recognition system, which includes: a parameter acquisition module, a factor calculation and model derivation module, and a real-time data processing and evaluation output module.

[0136] The parameter acquisition module is used to acquire target training parameters and real-time training parameters.

[0137] The factor calculation and model derivation module is used to calculate the training density and the intermittent and intensity coupling factor according to the target training parameters, process the training density and the intermittent and intensity coupling factor by using a random forest model, output a dynamic compensation coefficient and an initial intensity correction gradient, and perform a non-linear constraint on the intensity correction gradient.

[0138] The real-time data processing and evaluation output module is used to process the dynamic compensation coefficient, real-time training parameters and intensity correction gradient by using an attention gating network, including: the real-time training parameters include the actual inter-group interval deviation rate, and the STL decomposition method is used to perform time series decomposition on the actual inter-group interval deviation rate to generate a trend term and a mutation term; the weights of the attention heads are assigned by the dynamic compensation coefficient and the intensity correction gradient, and the multi-head attention mechanism is used to enhance the features of the trend term to generate a trend enhancement term; the training efficiency index and the anomaly risk level are output according to the stability of the trend enhancement term and the peak value of the mutation term.

[0139] This embodiment has all the advantages of the intelligent monitoring and anomaly recognition method for strength training, and can automatically execute the steps of the intelligent monitoring and anomaly recognition method for strength training.

[0140] In some embodiments, the present disclosure provides an intelligent monitoring and anomaly recognition device for strength training, which includes a memory and a processor; the memory is used to store a computer program; the processor is used to implement the steps of the intelligent monitoring and anomaly recognition method for strength training when executing the computer program.

[0141] In some embodiments, the present disclosure provides a storage medium, in which computer program instructions are stored. When the computer program instructions are read and run by a processor, the steps of the intelligent monitoring and anomaly recognition method for strength training are executed.

[0142] Among them, any reference to a memory, storage, database or other medium used in the embodiments provided by the present invention may include non-volatile and / or volatile memories. The non-volatile memory may include a read-only memory (ROM), a programmable ROM (PROM), an electrically programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM) or a flash memory. The volatile memory may include a random access memory (RAM) or an external cache memory.

[0143] It should be noted that in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, method, article or device including the element.

[0144] Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent monitoring and abnormal recognition method for strength training, characterized in that, Including: Obtain target training parameters and real-time training parameters; Calculate the training density and the interval and intensity coupling factor according to the target training parameters, process the training density and the interval and intensity coupling factor using a random forest model, output the dynamic compensation coefficient and the intensity correction gradient, and perform a non-linear constraint on the intensity correction gradient; Process the dynamic compensation coefficient, real-time training parameters, and intensity correction gradient using an attention gating network, including: The real-time training parameters include the actual inter-group interval deviation rate. Use the STL decomposition method to perform time series decomposition on the actual inter-group interval deviation rate to generate a trend term and a mutation term; Allocate the weights of the attention heads through the dynamic compensation coefficient and the intensity correction gradient, and use the multi-head attention mechanism to enhance the features of the trend term to generate a trend enhancement term; Output the training efficiency index and the abnormal risk level according to the stability of the trend enhancement term and the peak value of the mutation term.

2. The power training intelligent monitoring and anomaly recognition method according to claim 1, wherein The target training parameters include the training volume, training intensity, target inter-group interval time, and target intra-group interval time; Calculate the training density and the interval and intensity coupling factor according to the target training parameters, including: Divide the training volume by the product of the target inter-group interval time and the target intra-group interval time to obtain the training density; Multiply the difference between the target inter-group interval time and the target intra-group interval time by the training intensity to obtain the interval and intensity coupling factor.

3. The power training intelligent monitoring and anomaly recognition method according to claim 2, characterized in that, Process the training density and the interval and intensity coupling factor using a random forest model, and output the dynamic compensation coefficient and the intensity correction gradient, including: Normalize the training density and the interval and intensity coupling factor as the dynamic feature weights of each decision tree; Based on the dynamic feature weights, use the CART algorithm for node splitting to generate the dynamic compensation coefficient and the intensity correction gradient.

4. The method for intelligent monitoring and abnormal recognition of strength training according to claim 3, wherein The real-time training parameters also include the load weight, and calculate the load volatility according to the load weight; Perform a non-linear constraint on the intensity correction gradient, including: Set the base range, the first threshold, and the lowest upper limit value of the intensity correction gradient. When the load volatility exceeds the first threshold, reduce the upper limit value of the base range to the lowest upper limit value; Perform non-linear compression after scaling the intensity correction gradient with the actual inter-group interval deviation rate; Use the gradient clipping algorithm to perform boundary constraint on the compressed intensity correction gradient based on the base range.

5. The method for intelligent monitoring and abnormal recognition of strength training according to claim 4, wherein Divide the difference between the load weight and the training intensity by the training intensity to obtain the load volatility.

6. The method for intelligent monitoring and abnormal recognition of strength training according to claim 3, characterized in that The real-time training parameters also include the actual intra-group interval deviation rate; Use the STL decomposition method to perform time series decomposition on the actual inter-group interval deviation rate to generate a trend term and a mutation term, including: Filter out outliers and fill in missing values for the actual inter-group interval deviation rate to generate time series data; Determine the decomposition window length according to the length of the time series data, and perform smoothing processing on the data within each window to output the trend term; Use the difference between the time series data and the trend term as the mutation term, and adjust the mutation term amplitude based on the dynamic compensation coefficient.

7. The strength training intelligent monitoring and anomaly recognition method according to claim 6, characterized in that Allocate the weights of the attention heads through the dynamic compensation coefficient and the intensity correction gradient, and use the multi-head attention mechanism to enhance the features of the trend term to generate a trend enhancement term, including: Perform weighted summation on the dynamic compensation coefficient and the intensity correction gradient to generate the allocation weights of each attention head in the multi-head attention mechanism; The multi-head attention mechanism is adopted to enhance the features of the trend term, and the output results of multiple attention heads are normalized and weighted and fused to generate a trend enhancement term.

8. The strength training intelligent monitoring and anomaly recognition method according to claim 3, wherein The training efficiency index and the abnormal risk level are output according to the stability of the trend enhancement term and the peak value of the mutation term, including: The trend enhancement term is divided into windows, and the variance of all trend enhancement terms included in each window is calculated as the stability index; The maximum absolute value of the mutation term in the time series is extracted as the peak value; Weights are assigned to the stability index and the peak value, and the weighted sum result of the stability index and the peak value amplitude is linearly mapped to a specified interval to obtain the training efficiency index; the abnormal risk level is output according to the peak value and the dynamic compensation coefficient.

9. An intelligent monitoring and abnormal recognition system for strength training, characterized in that, It includes: A parameter acquisition module, which is used to acquire target training parameters and real-time training parameters; A factor calculation and model derivation module, which is used to calculate the training density and the intermittent and intensity coupling factor according to the target training parameters, process the training density and the intermittent and intensity coupling factor by using a random forest model, output the dynamic compensation coefficient and the initial intensity correction gradient, and perform non-linear constraint on the intensity correction gradient; A real-time data processing and evaluation output module; It is used to process the dynamic compensation coefficient, the real-time training parameters and the intensity correction gradient by using an attention gating network, including: The real-time training parameters include the actual inter-group interval deviation rate. The STL decomposition method is used to perform time series decomposition on the actual inter-group interval deviation rate to generate a trend term and a mutation term; The weights of the attention heads are assigned by the dynamic compensation coefficient and the intensity correction gradient, and the multi-head attention mechanism is used to enhance the features of the trend term to generate a trend enhancement term; The training efficiency index and the abnormal risk level are output according to the stability of the trend enhancement term and the peak value of the mutation term.

10. A storage medium, characterized in that, Computer program instructions are stored in a storage medium. When the computer program instructions are read and run by a processor, the steps of the strength training intelligent monitoring and abnormal recognition method described in any one of claims 1-8 are executed.

Images