Surface electromyogram signal denoising method based on crown porcupine improved variational mode decomposition

Abstract

The invention relates to the field of surface electromyogram signal denoising, in particular to a surface electromyogram signal denoising method based on crown porcupine improved variational mode decomposition. The method comprises the steps that firstly, key parameters of VMD are initialized, self-adaptive decomposition is conducted on collected original surface electromyogram signals, a set of intrinsic mode functions IMFs are obtained, and the key parameters comprise a mode number K and a penalty factor alpha; then, parameter selection of VMD is optimized in combination with a crown porcupine optimization algorithm (CPO), sEMG signals are accurately decomposed into a plurality of independent frequency bands, finally, IMFs of different frequency bands are screened and reconstructed, final pure surface electromyogram signals are obtained, muscle activities in the signals can be clearly distinguished, and therefore the denoising effect of the surface electromyogram signals is improved; and signals with higher quality are provided for signal analysis in the subsequent rehabilitation process.

Description

Technical Field

[0001] The invention relates to the field of surface electromyographic signal denoising, and in particular to a surface electromyographic signal denoising method based on improved variational mode decomposition of crested porcupine. Background Art

[0002] According to the China Stroke Prevention and Treatment Report, one person in my country suffers from a primary or recurrent stroke every ten seconds, and approximately 50%-70% of stroke patients will experience limb dysfunction, so rehabilitation technology is essential to maximize efficacy and reduce disability. However, during the rehabilitation process, the evaluation of limb function recovery often relies on the empirical judgment of rehabilitation physicians, resulting in insufficient accuracy and personalization of rehabilitation effects. To overcome this problem, in recent years, objective motor function detection methods have gradually been introduced into rehabilitation medicine, among which motion recognition technology based on surface electromyography (sEMG) has received widespread attention due to its non-invasiveness, high correlation, and real-time monitoring capabilities.

[0003] Surface electromyography (sEMG) is a bioelectric signal that records neuromuscular activity on the surface of the skin through electromyography sensors. It can provide rich information about the patient's muscle activity and movement patterns. Using sEMG signals to analyze the different activity patterns of patients' muscles can help rehabilitation physicians evaluate patients' athletic performance in real time and optimize training programs. However, since sEMG signals are easily affected by external interference (such as motion artifacts, electromagnetic noise, and equipment vibration), their signal quality and the accuracy of motion recognition are often severely restricted. Noise interference can cause distortion of the frequency domain characteristics of sEMG signals, resulting in a decrease in the classification accuracy of the motion recognition model and even misjudgment. Therefore, ensuring effective denoising of sEMG before signal analysis is a key link in improving the accuracy of analysis of different muscle activity patterns of patients and improving rehabilitation effects.

[0004] Empirical mode decomposition is a very suitable method for processing nonlinear and non-stationary signals, but it has problems such as fitting overshoot, endpoint effect and mode aliasing, which limit its practical application. The traditional variational mode decomposition method relies on experience to set relevant parameters, which may lead to poor decomposition effect and sEMG signal noise removal performance. In the parameter selection process of VMD, traditional adaptive algorithms such as particle swarm optimization, genetic algorithm, ant colony algorithm, etc. can achieve certain results in optimization, but in combination with VMD method, its local optimal problem is still difficult to solve. Summary of the invention

[0005] In order to solve the problem of low denoising accuracy of surface EMG signals, the present invention provides a surface EMG signal denoising method based on modified variational modal decomposition of crested porcupine. The method combines the crested porcupine optimization algorithm (CPO) to optimize the parameter selection of VMD, and accurately decomposes the sEMG signal into multiple independent frequency bands, so that the muscle activity in the signal can be clearly distinguished, thereby improving the denoising effect of the surface EMG signal and providing higher quality signals for signal analysis in the subsequent rehabilitation process.

[0006] A surface electromyography signal denoising method based on modified variational mode decomposition of crested porcupine mainly includes the following steps:

[0007] S1: Initialize VMD parameters, adaptively decompose the collected original surface electromyographic signals, and obtain a set of intrinsic mode functions IMFs. The VMD parameters include the mode number K and the penalty factor α.

[0008] S2: Use CPO to optimize VMD parameters. CPO simulates the four defense strategies of crested porcupines to perform global search in the search space to find the optimal solution for the combination of VMD parameters. In each iteration, CPO calculates the fitness of each individual according to the fitness function, uses these fitness values ​​to guide the search process, updates the individual positions, and gradually optimizes the VMD parameters until the convergence condition is reached.

[0009] S3: According to the optimal VMD parameter combination, VMD is optimized, and the optimized variational mode decomposition method is used to optimally adaptively decompose the sEMG signal to obtain IMFs in different frequency bands;

[0010] S4: Calculate the energy of IMFs in different frequency bands, select IMFs whose energy exceeds the preset energy threshold as valid electromyographic signals, and then add and reconstruct the selected IMFs to obtain the final pure surface electromyographic signals.

[0011] A computer device comprises a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above method.

[0012] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above method.

[0013] A computer program product comprises a computer program or instructions, and when the program or instructions are executed by a processor, the steps of the above method are implemented.

[0014] The beneficial effects brought by the technical solution provided by the present invention are:

[0015] 1. This invention combines the CPO optimization algorithm to automatically optimize the key parameters of VMD (modal number K and penalty factor α), avoiding the limitations of manual parameter selection in traditional methods, thereby significantly improving the accuracy and efficiency of signal decomposition. This optimization method can ensure that signal decomposition is more accurate and scientific in a complex noise environment, and improve the denoising effect;

[0016] 2. Through the global search and local optimization mechanism of CPO, the problem that traditional adaptive algorithms are prone to fall into local optimality when combined with VMD is solved, so as to accurately select the number of VMD modes and penalty factors and accurately decompose different frequency bands in sEMG signals. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0018] Figure 1 It is a flow chart of a surface electromyography signal denoising method based on improved variational mode decomposition of crested porcupine in an embodiment of the present invention;

[0019] Figure 2 is a schematic diagram of the defense mechanism of CPO simulating crested porcupine in an embodiment of the present invention;

[0020] Figure 3 Schematic diagram of the method proposed in the embodiment of the present invention accurately decomposing the surface electromyographic signal into IMFs of a single frequency. DETAILED DESCRIPTION

[0021] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, specific embodiments of the present invention are now described in detail with reference to the accompanying drawings.

[0022] Example 1

[0023] Please refer to Figure 1 , Figure 1It is a flow chart of a surface electromyography signal denoising method based on the variational modal decomposition improved by crown porcupine in an embodiment of the present invention. The method obtains high-quality surface electromyography signals by optimizing the key parameters of the variational modal decomposition (VMD) method. VMD is selected to be used. VMD can effectively process non-stationary signals, avoid the common mode aliasing problem in traditional methods, and has strong noise suppression ability. sEMG signals are usually affected by noise such as power line interference and motion artifacts, and VMD can effectively separate useful signals from noise by adaptively decomposing the signal into several band-limited modes. In addition, the Wiener filtering characteristics of VMD enable it to adaptively filter according to the frequency characteristics of the signal, thereby improving the denoising effect. Although VMD relies on key parameters (such as the number of modes K and the penalty factor α) during the decomposition process, the optimization of these parameters usually needs to be achieved through intelligent algorithms to improve the decomposition accuracy. The high efficiency and time-frequency analysis capabilities of VMD enable it to provide more detailed signal analysis when processing complex sEMG signals, and is particularly suitable for applications such as rehabilitation training and muscle fatigue monitoring.

[0024] The method specifically comprises the following steps:

[0025] S1: Initialize VMD parameters, adaptively decompose the collected original surface electromyographic signals, and obtain a set of intrinsic mode functions IMFs. The VMD parameters include the mode number K and the penalty factor α.

[0026] S2: Use CPO to optimize VMD parameters. CPO simulates the four defense strategies of crested porcupines to perform global search in the search space to find the optimal solution for the global VMD parameter combination. In each iteration, CPO calculates the fitness of each individual according to the fitness function. CPO uses these fitness values ​​to guide the search process, update the individual position, and gradually optimize the VMD parameters until the convergence condition is reached. The fitness function uses information entropy to measure the quality of each signal decomposition result. The process of CPO calculating the fitness of each individual according to the fitness function is: according to the IMFs obtained after signal decomposition and the noise removal effect, the fitness of each solution is calculated. A lower fitness value means less signal noise and lower information entropy. Information entropy is used to quantify the disorder or uncertainty of the signal during the optimization process, which can effectively evaluate the quality of each solution (i.e., each individual) in each iteration.

[0027] At the beginning of the CPO algorithm, the positions of multiple individuals are initialized. These positions represent different combinations of VMD parameters. Each individual decomposes the signal according to its corresponding VMD parameters to obtain a set of IMFs. Then, based on the decomposition results, the fitness of each individual is calculated using information entropy. If an individual can successfully reduce the noise in the signal and retain valid information, its fitness value will be higher.

[0028] S3: According to the optimal VMD parameter combination, VMD is optimized, and the optimized variational mode decomposition method is used to optimally adaptively decompose the sEMG signal to obtain IMFs in different frequency bands;

[0029] S4: Calculate the energy of IMFs in different frequency bands and select those IMFs with higher energy, that is, select IMFs with energy exceeding a preset energy threshold as valid electromyographic signals. Then, add and reconstruct the selected IMFs to obtain the final pure surface electromyographic signals.

[0030] The process of screening and reconstruction is to measure the effectiveness of each IMF by calculating its energy. IMFs with higher energy usually represent the main components of the signal, while IMFs with lower energy may be noise. IMFs with higher energy are screened out based on their energy values, and IMFs with energy above a certain threshold are usually retained. The screened IMFs are then added and reconstructed.

[0031] In order to adaptively optimize VMD parameters, the crested porcupine optimization algorithm is used to optimize VMD parameter selection. The crested porcupine optimization algorithm (CPO) simulates the defense mechanism of the crested porcupine, such as Figure 2 As shown in the figure, CPO simulates four defense strategies of crested porcupines: visual, acoustic, odor, and physical attack. These strategies range from the least aggressive to the most aggressive. The first and second defense strategies (i.e., visual and acoustic) represent the exploratory behavior of CPO, while the third and fourth defense strategies (i.e., odor and physical attack) represent the exploitative behavior of CPO. When the predator is far away, crested porcupines have two defense strategies, namely, visual strategy and acoustic strategy. These strategies refer to investigating different areas and devoting themselves to comprehensive exploration search. When the predator approaches, crested porcupines initiate the third and fourth defense strategies, namely, odor and physical attack. These two strategies refer to the development of potential areas and devoting themselves to local development search to approach the optimal solution. CPO achieves a balance between global search and local optimization by combining the exploration phase and the exploitation phase of crested porcupines. First, the algorithm extensively searches for possible solutions of the number of modalities and penalty factors through the initial exploration behavior, ensuring that the best combination of parameters is explored over a large range. During the preliminary search process, the population contraction mechanism of CPO is used to maintain the diversity of the population and prevent the algorithm from falling into the local optimum too early. Subsequently, the CPO algorithm entered the development phase, by refining the search range, focusing on the parameter regions with better performance, further optimizing the number of modes and penalty factors, and ensuring that VMD can effectively decompose the signal. The better parameter region is the parameter combination region with better VMD decomposition effect. The better standard can set a threshold as needed, and the parameter combination exceeding this threshold is regarded as the better parameter region. The VMD parameters corresponding to these regions can reduce the impact of noise and retain effective signal components. The comparison standard is the size of the fitness value, that is, the size of the information entropy.

[0032] Compared with the traditional EMD method, VMD can effectively avoid the "mode aliasing" problem. In the decomposition process, the EMD method often mixes signals of multiple frequency bands in the same mode, resulting in inaccurate decomposition results. VMD optimizes the algorithm to ensure that each mode represents an independent frequency band, thereby reducing confusion in signal decomposition and making each mode more accurately represent a certain frequency component of the signal, such as Figure 3 As shown, the method proposed in the present invention accurately decomposes the surface electromyographic signal into IMFs of a single frequency.

[0033] The present invention applies the CPO optimization algorithm to the parameter optimization of VMD, adaptively searches for the optimal parameter combination of VMD, avoids the problem of traditional methods relying on experience to select parameters, improves the denoising effect and signal decomposition accuracy, and enables VMD to achieve the best denoising effect when processing actual complex sEMG signals.

[0034] Example 2

[0035] A computer device comprises a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above method.

[0036] Example 3

[0037] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above method.

[0038] Example 4

[0039] A computer program product comprises a computer program or instructions, and when the program or instructions are executed by a processor, the steps of the above method are implemented.

[0040] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A surface electromyography signal denoising method based on modified variational mode decomposition of crested porcupine, characterized in that: The steps include: S1: Initialize VMD parameters, adaptively decompose the collected original surface electromyographic signals, and obtain a set of intrinsic mode functions IMFs. The VMD parameters include the mode number K and the penalty factor α. S2: Use CPO to optimize VMD parameters. CPO simulates the four defense strategies of crested porcupines to perform global search in the search space to find the optimal solution for VMD parameter combination; In each iteration, CPO calculates the fitness of each individual according to the fitness function, uses these fitness values ​​to guide the search process, updates the individual positions, and gradually optimizes the VMD parameters until the convergence condition is reached; S3: According to the optimal VMD parameter combination, VMD is optimized, and the optimized variational mode decomposition method is used to optimally adaptively decompose the sEMG signal to obtain IMFs in different frequency bands; S4: Calculate the energy of IMFs in different frequency bands, select IMFs whose energy exceeds the preset energy threshold as valid electromyographic signals, and then add and reconstruct the selected IMFs to obtain the final pure surface electromyographic signals.

2. The surface electromyography signal denoising method based on the improved variational mode decomposition of crested porcupine as claimed in claim 1 is characterized in that: In S2, the fitness function uses information entropy to measure the quality of each signal decomposition result; the fitness of each solution is calculated based on the IMFs and noise removal effect obtained after signal decomposition.

3. The surface electromyography signal denoising method based on the modified variational mode decomposition of crested porcupine as claimed in claim 1, characterized in that: In S2, CPO simulated four defense strategy behaviors of crested porcupines: visual, sound, odor, and physical attack. The visual and sound defense strategies represent the exploratory behavior of CPO, while the odor and physical attack defense strategies represent the exploitative behavior of CPO.

4. The surface electromyography signal denoising method based on the modified variational mode decomposition of crested porcupine as claimed in claim 3 is characterized in that: When the predator is far away, the crested porcupine activates visual and sound strategies to investigate different areas and is committed to comprehensive exploration and search; when the predator is close, the crested porcupine activates scent and physical attacks to develop potential areas and is committed to local development and search to approach the optimal solution.

5. The surface electromyography signal denoising method based on the improved variational mode decomposition of crested porcupine as claimed in claim 4 is characterized in that: CPO achieves a balance between global search and local optimization by combining the exploration phase and development phase of the crested porcupine. The specific process is as follows: (1) The CPO algorithm searches extensively for possible solutions of the modal number K and the penalty factor α through initial exploration behavior, ensuring that the optimal combination of parameters is explored over a wide range; (2) Based on step (1), the CPO algorithm enters the development stage. By refining the search range and focusing on the parameter region with better performance, the mode number K and the penalty factor α are further optimized to ensure that VMD can effectively decompose the signal.

6. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the surface electromyography signal denoising method based on improved variational mode decomposition of crested porcupine as described in any one of claims 1 to 5.

7. A computer-readable storage medium, characterized in that: A computer program is stored, and when the program is executed by a processor, the steps of the surface electromyography signal denoising method based on improved variational mode decomposition of crested porcupines as described in any one of claims 1 to 5 are implemented.

8. A computer program product, characterized in that The method comprises a computer program or an instruction, which, when executed by a processor, implements the steps of the surface electromyography signal denoising method based on improved variational mode decomposition of crested porcupine as described in any one of claims 1 to 5.

Images