A muscle fatigue prediction method and system based on BP neural network model

Abstract

The present invention relates to the technical field of muscle fatigue prediction, and more particularly to a muscle fatigue prediction method and system based on a BP neural network model, comprising: filtering raw electromyographic signal data to obtain electromyographic signal data; obtaining standard electromyographic signal data for each individual based on the maximum isometric contraction corresponding to each individual and the root mean square value of each individual; and obtaining the integrated electromyographic value and wavelet packet entropy of each muscle; optimizing the BP neural network configuration to obtain an optimized BP neural network structure; training the optimized BP neural network structure based on the integrated electromyographic value, wavelet packet entropy, electromyographic signal data set, and fatigue score results of each muscle to obtain a trained BP neural network structure; and predicting muscle fatigue using the trained BP neural network structure. The present invention improves the accuracy of muscle fatigue prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of muscle fatigue prediction, and in particular to a muscle fatigue prediction method and system based on a BP neural network model. Background Art

[0002] Traditional muscle fatigue prediction and analysis often relies on subjective scoring methods. However, these methods (e.g., questionnaires) lack objectivity and real-time performance, making it difficult to accurately quantify muscle fatigue during dynamic tasks (e.g., overhead maintenance). Furthermore, surface electromyography (SEMG) signals, as one-dimensional time series, are subject to significant noise interference and individual differences, resulting in noisy raw data that cannot be directly used as model input. Finally, traditional BP (backpropagation) neural networks are sensitive to initial weight and threshold settings, making them prone to local optimality or slow convergence, impacting prediction accuracy. Summary of the Invention

[0003] The present invention provides a muscle fatigue prediction method and system based on a BP neural network model, which is used to solve the problems of subjectivity, poor real-time performance, large noise interference, non-standard data, and insufficient accuracy caused by the convergence problem of the traditional BP neural network in traditional muscle fatigue prediction methods.

[0004] The purpose of the present invention can be achieved through the following technical solutions:

[0005] The first aspect of the present invention is to provide a muscle fatigue prediction method based on a BP neural network model, comprising:

[0006] Determine the range of individual sample sizes for electromyographic testing;

[0007] Acquire several groups of raw electromyographic signal data from the subjects; filter the raw electromyographic signal data to obtain final filtered signal data, which are recorded as electromyographic signal data; obtain the maximum isometric contraction corresponding to each individual and the root mean square value of each individual in the electromyographic signal data; obtain the standardized electromyographic signal data of each individual based on the maximum isometric contraction corresponding to each individual and the root mean square value of each individual; obtain the integrated electromyographic value and wavelet packet entropy of each muscle based on the standardized electromyographic signal data of each individual;

[0008] Optimizing the BP neural network configuration to obtain an optimized BP neural network structure; obtaining a fatigue score result for each muscle; determining an electromyographic signal data set through a range of individual sample sizes for electromyographic physiological testing; training the optimized BP neural network structure based on the integrated electromyographic value of each muscle, wavelet packet entropy, the electromyographic signal data set, and the fatigue score result for each muscle to obtain a trained BP neural network structure;

[0009] Muscle fatigue is predicted through the trained BP neural network structure.

[0010] Furthermore, determining the range of the individual sample size for electromyography testing includes:

[0011]

[0012] Where, The degrees of freedom represent the probability of rejecting the true value, represents the degrees of freedom for the probability of false positives, Represents the standard deviation of the integrated EMG value of the muscle electrical signal, Indicates the basic value of the individual sample size for surface electromyography testing, It represents the difference in the integrated EMG value between the experimental group and the control group. Express Round up;

[0013] The range of the individual sample size of the electromyography test is determined by the basic value of the individual sample size of the surface electromyography test, that is, the individual sample size of the electromyography test must be greater than or equal to the basic value of the individual sample size of the surface electromyography test.

[0014] Furthermore, the original electromyographic signal data is filtered to obtain final filtered signal data, which is recorded as electromyographic signal data, including:

[0015] The original electromyographic signal data is first filtered using a bandpass filtering algorithm, then filtered using a notch filtering algorithm, and finally filtered using a wavelet filtering algorithm; the final filtered signal data is recorded as electromyographic signal data.

[0016] Furthermore, the step of obtaining the maximum isometric contraction corresponding to each individual and the root mean square value of each individual in the electromyographic signal data; obtaining the standard electromyographic signal data of each individual according to the maximum isometric contraction corresponding to each individual and the root mean square value of each individual, includes:

[0017] The maximum isometric contraction of each individual was obtained using a force sensor;

[0018] At preset equal intervals , obtain data of several moments in the electromyographic signal of each individual;

[0019]

[0020] Where, Represents the first Data at a moment, Represents the number of data at all times in the electromyographic signal of each individual, represents the root mean square value of each individual.

[0021] Furthermore, the step of obtaining the integrated electromyographic value and wavelet packet entropy of each muscle based on the standard electromyographic signal data of each individual includes:

[0022] According to the standard electromyographic signal data of each individual, the integrated electromyographic value of each muscle is obtained by integral calculation; according to the standard electromyographic signal data of each individual, the wavelet packet entropy of each muscle is obtained by wavelet packet transform algorithm.

[0023] Furthermore, the optimized BP neural network configuration obtains an optimized BP neural network structure; and obtains the fatigue score result of each muscle, including:

[0024] The optimized BP neural network structure is obtained by optimizing the BP neural network configuration through genetic algorithm;

[0025] Among them, the fatigue score of each muscle is evaluated manually.

[0026] Furthermore, the muscle fatigue prediction using the trained BP neural network structure includes:

[0027] The EMG signal data is used as input, and the trained BP neural network structure outputs the muscle fatigue score. The muscle fatigue score is stored in 10 values, ranging from 0 to 10. Among them, 0 to 3 indicates no fatigue, 4 to 6 indicates some fatigue, and 6 or above indicates very fatigue.

[0028] The muscle fatigue score result is used as the muscle fatigue prediction result.

[0029] The second aspect of the present invention is to provide a muscle fatigue prediction system based on a BP neural network model, comprising:

[0030] Sample range determination module: used to determine the range of individual sample size for electromyography physiological testing;

[0031] Data processing module: used to obtain several groups of raw electromyographic signal data from the subjects; filter the raw electromyographic signal data to obtain final filtered signal data, which is recorded as electromyographic signal data; obtain the maximum isometric contraction corresponding to each individual and the root mean square value of each individual in the electromyographic signal data; obtain the standard electromyographic signal data of each individual based on the maximum isometric contraction corresponding to each individual and the root mean square value of each individual; obtain the integrated electromyographic value and wavelet packet entropy of each muscle based on the standard electromyographic signal data of each individual;

[0032] Neural network optimization and training module: used to optimize the BP neural network configuration and obtain the optimized BP neural network structure; obtain the fatigue score results of each muscle; determine the electromyographic signal data set through the range of individual sample size of electromyographic physiological tests; train the optimized BP neural network structure based on the integrated electromyographic value of each muscle, wavelet packet entropy, electromyographic signal data set and the fatigue score results of each muscle to obtain the trained BP neural network structure;

[0033] Prediction module: used to predict muscle fatigue through the trained BP neural network structure.

[0034] The third aspect of the present invention is to provide an electronic device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor, wherein the processor implements the muscle fatigue prediction method based on the BP neural network model when executing the computer program.

[0035] A fourth aspect of the present invention is to provide a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the muscle fatigue prediction method based on the BP neural network model is implemented.

[0036] Compared with the prior art, the present invention has the following beneficial effects: filtering the original electromyographic signal data to obtain electromyographic signal data, reducing the influence of noise on the electromyographic signal data; obtaining the maximum isometric contraction corresponding to each individual and the root mean square value of each individual in the electromyographic signal data, reducing the influence of individual differences; obtaining the standard electromyographic signal data of each individual based on the maximum isometric contraction corresponding to each individual and the root mean square value of each individual; obtaining the integrated electromyographic value and wavelet packet entropy of each muscle based on the standard electromyographic signal data of each individual, thereby improving the calculation speed; optimizing the BP neural network configuration to obtain the optimized BP neural network structure, thereby reducing the problem that the BP neural network is sensitive to initial weights and threshold settings and is prone to falling into local optimality or slow convergence speed; obtaining the fatigue score result of each muscle; determining the training set according to the range of the individual sample size of the electromyographic physiological test; training the optimized BP neural network structure according to the integrated electromyographic value, wavelet packet entropy, electromyographic signal data set and the fatigue score result of each muscle to obtain the trained BP neural network structure; predicting muscle fatigue through the trained BP neural network structure, thereby improving the accuracy of muscle fatigue prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0038] Figure 1 The present invention provides a schematic flow chart of the steps of a muscle fatigue prediction method based on a BP neural network model;

[0039] Figure 2 The present invention provides a module flow diagram of a muscle fatigue prediction system based on a BP neural network model. DETAILED DESCRIPTION

[0040] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0041] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0042] In response to the problems existing in the background technology, a muscle fatigue prediction method and system based on the BP neural network model was studied and designed, which has important practical significance.

[0043] like Figure 1 As shown, the first aspect of the present invention is to provide a muscle fatigue prediction method based on a BP neural network model, comprising the following steps:

[0044] Step S001: Determine the range of individual sample size for electromyography testing.

[0045] It should be noted that surface electromyography (EMG) signals are one-dimensional time series signals obtained by amplifying, guiding, displaying, and recording the bioelectrical changes generated by neuromuscular activity through the skin surface. They can reflect the true degree of muscle fatigue in the human body and are characterized by non-invasive, real-time, multi-targeted, and fixed-point monitoring, as well as being unaffected by posture and environmental factors. Targeting typical upper limb muscles, a multi-channel physiological multiconductor was used to test the responses of different muscles to the same maintenance maneuver. The time and frequency domain signal values ​​were collected and processed, and the relationships between the data were analyzed to study the fatigue characteristics of the upper limbs during maintenance operations.

[0046] Specifically, the basic value of the surface electromyography test individual sample size is estimated according to the random design individual sample size estimation formula based on t distribution in statistics; wherein the random design individual sample size estimation formula is specifically expressed as:

[0047]

[0048] Where, The degrees of freedom represent the probability of rejecting the true value, represents the degrees of freedom for the probability of false positives, Represents the standard deviation of the integrated EMG value of the muscle electrical signal, Indicates the basic value of the individual sample size for surface electromyography testing, It represents the difference in the integrated EMG value between the experimental group and the control group. Express Round up. Among them, the standard deviation of the integrated electromyographic value of the muscle electrical signal The difference in integrated EMG values ​​between the experimental group and the control group They are all determined through literature or expert experience in the field; among them, =0.05, .

[0049] In this embodiment, the probability of rejecting a true result is 0.05, and the probability of removing a false result is 0.1. In this embodiment, there is no specific limitation on the probability of rejecting a true result and the probability of removing a false result, and the implementer may determine them according to the specific situation.

[0050] in, The corresponding degrees of freedom when the probability of rejecting the true value is 0.05 is 1.645. The corresponding degrees of freedom when the probability of false positive is 0.1 is 1.282.

[0051] According to the random design individual sample size estimation formula, the basic value of the individual sample size for electromyography testing is:

[0052]

[0053] So far, the basic value of the individual sample size of electromyography test has been obtained through the above method; therefore, in the subsequent analysis, it must be ensured that the range of the individual sample size of electromyography test is greater than or equal to .

[0054] Step S002: Filter the original electromyographic signal data to obtain electromyographic signal data; obtain the maximum isometric contraction corresponding to each individual and the root mean square value of each individual in the electromyographic signal data; obtain the standard electromyographic signal data of each individual based on the maximum isometric contraction corresponding to each individual and the root mean square value of each individual; obtain the integrated electromyographic value and wavelet packet entropy of each muscle based on the standard electromyographic signal data of each individual.

[0055] It should be noted that, for the sake of rigor in the experimental design, eight channels were collected during the collection of EMG data, covering all major muscles of the upper limb. However, after collecting these data and conducting preliminary analysis, the researchers found that compared to the resting state and the data of the other limb, only channels 2, 3, and 4, namely the data of the right biceps, deltoid, and trapezius muscles, showed significant fluctuations and a high muscle contribution rate. This is consistent with the observation during the experiment that antenna array installers usually use their right hand (dominant hand) to perform almost all overhead operations. Therefore, in the subsequent model implementation process, in order to avoid the impact of noise in other channel data on the accuracy of the model and the impact of data volume on the model running speed, the EMG data of other channels were eliminated, and only the EMG data of the right biceps, deltoid, and trapezius muscles were used as EMG signals for subsequent analysis.

[0056] It should be further explained that, due to the interference of noise when collecting electromyographic signal data, the collected electromyographic signal has a certain deviation, so the collected electromyographic signal is first subjected to denoising and filtering processing by a filtering method.

[0057] Specifically, several groups of original electromyographic signal data of the subjects are obtained; the original electromyographic signal data are first filtered using a bandpass filtering algorithm, then filtered using a notch filtering algorithm, and finally filtered using a wavelet filtering algorithm; the final filtered signal data is recorded as electromyographic signal data.

[0058] The bandpass filtering algorithm, notch filtering algorithm, and wavelet filtering algorithm are all well-known technologies and will not be described in detail here. In this embodiment, the frequency range of the bandpass filtering algorithm is 50 to 480 Hz, the specific frequency of the notch filtering algorithm is 50 Hz, and the frequency range of the wavelet filtering algorithm is 50 to 450 Hz. In this embodiment, neither the frequency range nor the specific frequency is specifically limited, and the implementer may determine the specific frequency based on the specific situation.

[0059] So far, the electromyographic signal data has been obtained through the above method.

[0060] It should be noted that due to the large differences between different individuals, the signal values ​​corresponding to the electromyographic signals of different individuals under the same muscle fatigue are different. Therefore, there will be large errors when training the neural network model directly based on the individual electromyographic signals. Therefore, in order to eliminate the differences between different individuals, the maximum isometric contraction (MVC, Maximum Voluntary Contraction) corresponding to each individual is introduced to eliminate the difference problem.

[0061] Specifically, the maximum isometric contraction of each individual is obtained by a force sensor (force plate or tensiometer). , obtain data at several moments in the electromyographic signal of each individual; obtain the root mean square (RMS) value of each individual based on the data at several moments in the electromyographic signal of each individual; the root mean square value of each individual is specifically expressed by the formula:

[0062]

[0063] Where, Represents the first Data at a moment, Represents the number of data at all times in the electromyographic signal of each individual, Represents the root mean square value of each individual. In this embodiment, equal intervals are preset. Seconds, wherein, in this embodiment, the preset equal intervals There is no specific limitation and implementers can decide based on specific circumstances.

[0064] According to the root mean square value of each individual and the maximum isometric contraction of each individual, the standard electromyographic signal data of each individual is obtained; the standard electromyographic signal data of each individual is specifically expressed by the formula:

[0065]

[0066] Where, represents the root mean square value of each individual, represents the maximum isometric contraction of each individual, Represents the standard electromyographic signal data of each individual.

[0067] It should be noted that common characteristic indicators include the more traditional ones such as integrated electromyography (IEMG), average power frequency, median frequency, etc., and the newer ones include wavelet packet energy, energy proportion, energy entropy, etc.; in the early data processing, the researchers found that as the experiment progressed, the subjects' fatigue increased. Among the traditional indicators, IEMG (time domain electromyography signal) had a negative correlation trend, and among the new indicators, wavelet packet entropy (in the frequency domain) also had a strong change correlation; because IEMG has a strong correlation in the time domain, and wavelet packet entropy has a strong correlation in the frequency domain, the model simultaneously selected IEMG and wavelet packet entropy as secondary indicators of electromyography signals, so as to better characterize the fatigue change trend of the subjects during the entire experiment, and provide the neural network with more valuable learning features.

[0068] Specifically, based on the standard myoelectric signal data of each individual, the integrated myoelectric value of each muscle is obtained through integral calculation. Based on the standard myoelectric signal data of each individual, the wavelet packet entropy of each muscle is obtained through the wavelet packet transform algorithm. The calculation process of the integrated myoelectric value is well known in the art, and the process of calculating the wavelet packet entropy using the wavelet packet transform algorithm is also well known in the art and will not be further described here.

[0069] At this point, the integrated EMG value and wavelet packet entropy of each muscle are obtained through the above method.

[0070] Step S003: Optimize the BP neural network configuration to obtain the optimized BP neural network structure; train the optimized BP neural network structure according to the integrated electromyographic value of each muscle, wavelet packet entropy, electromyographic signal data set and the fatigue score of each muscle to obtain the trained BP neural network structure.

[0071] It should be noted that using a neural network to train and predict a large sample size with the number of experiments and the passage of time can produce prediction results with ten times the number of samples. This embodiment trains data based on a BP neural network model to obtain more reliable data results.

[0072] It should be further explained that the BP neural network includes an input layer, a hidden layer, and an output layer; the BP neural network is very sensitive to the initial connection weights between the input layer and the hidden layer neurons, and the initial thresholds between the hidden layer and the output layer. Therefore, unreasonable parameter settings will lead to slow network convergence and fall into local optimality.

[0073] Specifically, the BP neural network configuration is optimized using a GA (Genetic Algorithm) to obtain an optimized BP neural network structure. The optimization process obtains the optimal values ​​of all parameters in the BP neural network configuration. The process of optimizing the BP neural network configuration using a genetic algorithm is well known and will not be described in detail here.

[0074] The EMG signal dataset was determined based on the range of individual sample sizes from electromyographic physiological tests. The optimized BP neural network structure was trained based on the integrated EMG value, wavelet packet entropy, EMG signal dataset, and fatigue scores for each muscle. The trained BP neural network structure was obtained. The muscle fatigue scores were manually evaluated.

[0075] During the training process, the EMG signal dataset is divided into a training set and a test set in a ratio of 8:2. After multiple trainings and based on the training results, the network structure is improved and the hidden layer of the model is used. The model has a structure with three hidden layers. The first two layers use ReLU (Rectified Linear Unit) as the activation function, with a learning rate of 0.005. After 100 training epochs, the model can basically meet the requirements while maintaining a certain level of generalization. After the trained model parameters are stored as a weight file, the model training process is complete.

[0076] During the BP neural network training process, a computer equipped with an x86 processor (main frequency not less than 1GHz, memory not less than 1GB) and equipped with the Python 3.11.10 runtime library is required.

[0077] The third-party libraries called during the model implementation process include: Pytorch (neural network building framework), Pandas (external file reading, processing and writing), Openpyxl (Excel file reading support), TQDM (program execution progress display), Numpy (matrix operation support framework), Matplotlib (visual drawing support framework), Scipy (scientific operation and filter building library), Pywavelets (wavelet transform support library), OS (partial system interface library) and SYS (Python interpreter interaction support library).

[0078] Step S004: predicting muscle fatigue using the trained BP neural network structure.

[0079] The trained BP neural network structure uses electromyographic signal data as input and outputs a muscle fatigue score. The muscle fatigue score is stored on a 10-point scale, ranging from 0 to 10, with 0 to 3 indicating no fatigue, 3 to 6 indicating some fatigue, and 6 or higher indicating very fatigue. The muscle fatigue score is used as the muscle fatigue prediction result.

[0080] At this point, the muscle fatigue prediction is completed.

[0081] like Figure 2 As shown, the second aspect of the present invention is to provide a muscle fatigue prediction system based on a BP neural network model, comprising the following modules:

[0082] Sample range determination module 101: used to determine the range of individual sample size for electromyography test;

[0083] Data processing module 102: used to obtain several groups of raw electromyographic signal data from the subjects; filter the raw electromyographic signal data to obtain final filtered signal data, which is recorded as electromyographic signal data; obtain the maximum isometric contraction corresponding to each individual and the root mean square value of each individual in the electromyographic signal data; obtain the standardized electromyographic signal data of each individual based on the maximum isometric contraction corresponding to each individual and the root mean square value of each individual; and obtain the integrated electromyographic value and wavelet packet entropy of each muscle based on the standardized electromyographic signal data of each individual;

[0084] Neural network optimization and training module 103: for optimizing the BP neural network configuration to obtain an optimized BP neural network structure; obtaining a fatigue score result for each muscle; determining an electromyographic signal data set based on the range of individual sample size of electromyographic physiological tests; training the optimized BP neural network structure based on the integrated electromyographic value of each muscle, wavelet packet entropy, the electromyographic signal data set, and the fatigue score result of each muscle to obtain a trained BP neural network structure;

[0085] Prediction module 104: used to predict muscle fatigue through the trained BP neural network structure.

[0086] The third aspect of the present invention is to provide an electronic device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, a muscle fatigue prediction method based on a BP neural network model is implemented.

[0087] A fourth aspect of the present invention is to provide a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, a muscle fatigue prediction method based on a BP neural network model is implemented.

[0088] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, optical storage, etc.) containing computer-usable program code.

[0089] The present invention is described with reference to flowcharts and / or block diagrams of methods, systems, and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0090] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0091] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the present invention.

Claims

1. A muscle fatigue prediction method based on BP neural network model, characterized in that: include: Determine the range of individual sample sizes for electromyography testing, including: Where, The degrees of freedom represent the probability of rejecting the true value, represents the degrees of freedom for the probability of false positives, Represents the standard deviation of the integrated EMG value of the muscle electrical signal, Indicates the basic value of the individual sample size for surface electromyography testing, It represents the difference in the integrated EMG value between the experimental group and the control group. Express Round up; The range of the individual sample size of the electromyography test is determined by the basic value of the individual sample size of the surface electromyography test, that is, the individual sample size of the electromyography test must be greater than or equal to the basic value of the individual sample size of the surface electromyography test; Acquire several groups of raw electromyographic signal data from the subjects; filter the raw electromyographic signal data to obtain final filtered signal data, which is recorded as electromyographic signal data; obtain the maximum isometric contraction corresponding to each individual and the root mean square value of each individual in the electromyographic signal data; obtain standard electromyographic signal data for each individual based on the maximum isometric contraction corresponding to each individual and the root mean square value of each individual, including obtaining the maximum isometric contraction of each individual through a force sensor; At preset equal intervals , obtain data of several moments in the electromyographic signal of each individual; Where, Represents the first Data at a moment, Represents the number of data at all times in the electromyographic signal of each individual, represents the root mean square value of each individual; The standard electromyographic signal data of each individual is specifically expressed by the formula: Where, represents the maximum isometric contraction of each individual, Representing the standard electromyographic signal data of each individual; obtaining the integrated electromyographic value and wavelet packet entropy of each muscle according to the standard electromyographic signal data of each individual; Optimizing the BP neural network configuration to obtain an optimized BP neural network structure; obtaining a fatigue score result of each muscle, including optimizing the BP neural network configuration through a genetic algorithm to obtain an optimized BP neural network structure; wherein the fatigue score result of each muscle is manually evaluated; determining an electromyographic signal data set through the range of individual sample size of electromyographic physiological testing; training the optimized BP neural network structure based on the integrated electromyographic value of each muscle, wavelet packet entropy, the electromyographic signal data set, and the fatigue score result of each muscle to obtain a trained BP neural network structure; Muscle fatigue is predicted through the trained BP neural network structure.

2. The muscle fatigue prediction method based on the BP neural network model according to claim 1, characterized in that: The original electromyographic signal data is filtered to obtain final filtered signal data, which is recorded as electromyographic signal data, including: The original electromyographic signal data is first filtered using a bandpass filtering algorithm, then filtered using a notch filtering algorithm, and finally filtered using a wavelet filtering algorithm; the final filtered signal data is recorded as electromyographic signal data.

3. The muscle fatigue prediction method based on BP neural network model according to claim 1, characterized in that: The method of obtaining the integrated electromyographic value and wavelet packet entropy of each muscle based on the standard electromyographic signal data of each individual includes: According to the standard electromyographic signal data of each individual, the integrated electromyographic value of each muscle is obtained by integral calculation; according to the standard electromyographic signal data of each individual, the wavelet packet entropy of each muscle is obtained by wavelet packet transform algorithm.

4. The muscle fatigue prediction method based on the BP neural network model according to claim 1, characterized in that: The muscle fatigue prediction using the trained BP neural network structure includes: The EMG signal data is used as input, and the trained BP neural network structure outputs the muscle fatigue score. The muscle fatigue score is stored in 10 values, ranging from 0 to 10. Among them, 0 to 3 indicates no fatigue, 4 to 6 indicates some fatigue, and 6 or above indicates very fatigue. The muscle fatigue score result is used as the muscle fatigue prediction result.

5. A muscle fatigue prediction system based on BP neural network model, characterized in that: It includes a sample range determination module, a data processing module, a neural network optimization and training module, and a prediction module. When the sample range determination module, the data processing module, the neural network optimization and training module, and the prediction module are executed, a muscle fatigue prediction method based on a BP neural network model as described in claim 1 is implemented.

6. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the muscle fatigue prediction method based on the BP neural network model as described in any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for predicting muscle fatigue based on a BP neural network model according to any one of claims 1 to 4 is implemented.

Images