Muscle movement analysis method, device and equipment based on electromyogram data and storage medium

Abstract

The invention relates to the technical field of signal processing, and provides an electromyogram data-based muscle movement analysis method, device and equipment and a storage medium, the electromyogram data-based muscle movement analysis method comprises the following steps: using an adaptive filter to pre-process original electromyogram signals, and dynamically filtering noise signals; segmenting the electromyographic signal after the noise signal is filtered by using a sliding window, calculating the energy of the signal in each window, and detecting the zero-crossing rate of the signal in each window; performing wavelet threshold denoising on the original electromyographic signals to obtain wavelet domain features; fusing the signal energy, the zero-crossing rate and the wavelet domain features to form a feature vector; and inputting the feature vector into a pre-trained classification model for analysis to obtain a muscle motion state. The signal energy, the zero-crossing rate and the wavelet are fused with the features for analysis, real-time effective analysis of the muscle motion state is achieved, and the accuracy of muscle motion state analysis under the dynamic condition is improved.

Description

Technical Field

[0001] The present invention relates to the field of signal processing technology, and in particular to a muscle movement analysis method, device, equipment and storage medium based on electromyography data. Background Art

[0002] As a bioelectric signal reflecting muscle activity, electromyography data is widely used in rehabilitation medicine, motion monitoring, human-computer interaction and other fields. However, due to the unique characteristics of electromyography signals such as low amplitude, high noise and non-stationarity, conventional time domain analysis methods are difficult to fully and accurately reflect the dynamic activity of muscles, resulting in limited accuracy and real-time performance of signal analysis.

[0003] At present, frequency domain analysis is mostly used to extract electromyographic feature information, or simple time domain analysis is used to extract simple time domain features (such as mean, variance, etc.). It is difficult to effectively extract relevant information of complex muscle movements from low-amplitude, high-noise electromyographic signals. In addition, due to the lack of adaptability and real-time responsiveness in the data processing process, the accuracy of muscle movement state analysis in dynamic conditions is low. Summary of the invention

[0004] The embodiments of the present application provide a muscle movement analysis method, device, equipment and storage medium based on electromyography data, which realizes real-time and effective analysis of muscle movement state by analyzing signal energy, zero-crossing rate and wavelet and feature fusion, and improves the accuracy of muscle movement state analysis under dynamic conditions.

[0005] In a first aspect, an embodiment of the present application provides a muscle movement analysis method based on electromyography data, comprising:

[0006] Use adaptive filters to pre-process the original electromyographic signals and dynamically filter out noise signals;

[0007] The electromyographic signal after filtering out the noise signal is segmented using a sliding window, the energy of the signal in each window is calculated, and the zero-crossing rate of the signal in each window is detected;

[0008] Perform wavelet threshold denoising on the original electromyographic signal to obtain wavelet domain features;

[0009] The signal energy, zero-crossing rate and wavelet domain features are integrated to form a feature vector;

[0010] The feature vector is input into the pre-trained classification model for analysis to obtain the muscle movement state.

[0011] In a second aspect, an embodiment of the present application provides a muscle movement analysis device based on electromyography data, comprising:

[0012] The first denoising module is used to pre-process the original electromyographic signal using an adaptive filter to dynamically filter out noise signals;

[0013] A processing module, used for segmenting the electromyographic signal from which the noise signal is filtered out using a sliding window, calculating the energy of the signal in each window and detecting the zero-crossing rate of the signal in each window;

[0014] The second denoising module is used to perform wavelet threshold denoising on the original electromyographic signal to obtain wavelet domain features;

[0015] A construction module is used to fuse signal energy, zero-crossing rate and wavelet domain features to form a feature vector;

[0016] The analysis module is used to input the feature vector into a pre-trained classification model for analysis to obtain the muscle movement state.

[0017] In a third aspect, an embodiment of the present application provides an electronic device, including:

[0018] Memory and processing modules;

[0019] The memory is used to store computer programs;

[0020] The processing module is used to execute the computer program and implement the steps of the muscle movement analysis method based on electromyography data in the first aspect when executing the computer program.

[0021] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program;

[0022] When the computer program is executed by one or more processing modules, the one or more processing modules execute the steps of the muscle movement analysis method based on electromyography data in the first aspect as described above.

[0023] The embodiment of the present application provides a muscle movement analysis method, device, equipment and storage medium based on electromyographic data, wherein the muscle movement analysis method based on electromyographic data includes: using an adaptive filter to pre-process the original electromyographic signal and dynamically filter out the noise signal; using a sliding window to segment the electromyographic signal from which the noise signal is filtered, calculating the energy of the signal in each window and detecting the zero-crossing rate of the signal in each window; performing wavelet threshold denoising on the original electromyographic signal to obtain wavelet domain features; fusing the signal energy, zero-crossing rate and wavelet domain features to form a feature vector; inputting the feature vector into a pre-trained classification model for analysis to obtain the muscle movement state. By fusing the signal energy, zero-crossing rate and wavelet with the features for analysis, real-time and effective analysis of the muscle movement state is achieved, and the accuracy of the muscle movement state analysis under dynamic conditions is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0025] Figure 1 A schematic diagram of a muscle movement analysis method based on electromyography data provided in an embodiment of the present application;

[0026] Figure 2 A schematic diagram of the structure of a muscle movement analysis device based on electromyography data provided in an embodiment of the present application;

[0027] Figure 3 A schematic block diagram of a muscle movement analysis device based on electromyography data provided in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are 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 creative work are within the scope of protection of the present invention.

[0029] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0030] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0031] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0032] The technical solution provided by this application will be described in detail below with reference to the accompanying drawings.

[0033] See also Figure 1 , Figure 1A schematic diagram of a flow chart of a muscle movement analysis method based on electromyographic data provided in an embodiment of the present application. It should be noted that the muscle movement analysis method based on electromyographic data provided in an embodiment of the present application is performed by a muscle movement analysis device. The muscle movement analysis device includes but is not limited to a muscle rehabilitation physiotherapy instrument. Specifically, Figure 1 It can be seen that the muscle movement analysis method based on electromyography data includes steps S101 to S105. The details are as follows:

[0034] S101: Preprocess the original electromyographic signal using an adaptive filter to dynamically filter out noise signals.

[0035] The purpose of using adaptive filters to preprocess the original electromyographic signals (EMG signals) is to dynamically filter out noise signals, thereby obtaining clearer and more accurate electromyographic signal features. Electromyographic signals are often accompanied by various types of noise, such as power frequency interference, motion artifacts, ECG signal interference, etc., which directly affect the accuracy of subsequent analysis and classification. Adaptive filters can adjust their parameters in real time to adapt to changes in noise.

[0036] In this embodiment, the adaptive filter includes a weight adjustable filter and an adaptive algorithm. The weight adjustable filter is used to filter out a specific noise frequency band in the input signal, and the adaptive algorithm adjusts the filter weight according to the error signal between the filter output and the expected output, so that the filter can gradually adapt to the changes in the input signal. Specifically, the adaptive algorithm can be expressed as:

[0037] ω(n+1)=ω(n)+μ·e(n)·x(n)

[0038] Where ω(n) represents the filter weight at the nth iteration, μ is the step size factor that controls the adjustment speed, e(n) is the error signal that represents the difference between the filter output and the desired signal, and x(n) is the input signal.

[0039] S102: Segment the electromyographic signal from which the noise signal is filtered out using a sliding window, calculate the energy of the signal in each window, and detect the zero-crossing rate of the signal in each window.

[0040] Calculating the energy of the signal in each window includes: for any window, inputting the time domain signal value in the window into a preset energy calculation formula to calculate the energy of the signal in the window; wherein the preset energy calculation formula is:

[0041]

[0042] Among them, E(i) is the energy of the i-th window, reflecting the strength of muscle activity in the window; N is the length of the sliding window, and x(i+j) is the time domain signal value in the window. The energy feature can reflect the intensity of muscle activity. When the muscle is moving, the energy of the muscle electrical signal usually increases.

[0043] Detecting the zero-crossing rate of the signal in each window includes: for any window, inputting the time domain signal value in the window into a preset zero-crossing rate calculation formula to calculate the zero-crossing rate of the signal in the window; wherein the preset zero-crossing rate calculation formula is:

[0044]

[0045] Where sgn is the sign function, defined as:

[0046]

[0047] Z(i) represents the number of times the signal crosses zero in the i-th window.

[0048] From the sign function, we can see that when the signal changes from positive to negative or from negative to positive, the zero crossing rate increases. This feature is often used to distinguish different types of muscle movements.

[0049] S103: Perform wavelet threshold denoising on the original electromyographic signal to obtain wavelet domain features.

[0050] The original electromyographic signal is subjected to wavelet threshold denoising to obtain wavelet domain features, including: decomposing the original electromyographic signal into multiple frequency band signals based on wavelet transform, and denoising each frequency band signal based on a preset wavelet coefficient threshold; performing inverse wavelet transform on each frequency band signal after denoising based on the approximate coefficient and the detail coefficient after denoising, and extracting the wavelet domain features of each frequency band signal after inverse wavelet transform.

[0051] Wavelet transform transforms a signal into a set of wavelet coefficients through convolution with wavelet function. These coefficients represent the content of the signal at different time and frequency positions. Wavelet function has good localization characteristics in time and frequency domain and can effectively capture the instantaneous characteristics of the signal. For the original electromyographic signal, its wavelet transform can be expressed as:

[0052]

[0053] Among them, W x (a,b) is the wavelet coefficient, which represents the signal information at scale a and position b. a is the adjusted scale, b is the position, and x(t) is the original electromyographic signal. It is the mother wavelet function, which determines the characteristics of wavelet transform.

[0054] Wavelet transform can decompose the original electromyographic signal into components of different frequency bands by adjusting the scale a and position b. Generally speaking, a smaller scale a corresponds to high-frequency components and can capture rapid changes in the signal. A larger scale a corresponds to low-frequency components and reflects the overall trend or slow changes of the signal.

[0055] Exemplarily, the original electromyographic signal is decomposed into multiple frequency band signals based on wavelet transform, and each frequency band signal is denoised based on a preset wavelet coefficient threshold, including: performing multi-layer wavelet transform on the original electromyographic signal to decompose the original electromyographic signal into multiple frequency band signals, the wavelet coefficients of each frequency band signal include approximate coefficients and detail coefficients; and denoising the detail coefficient based on a preset wavelet coefficient threshold.

[0056] In wavelet transform, different wavelet coefficients W x (a,b) represents the components of the signal at a specific time position b and scale a. By decomposing the signal into multiple wavelet coefficients, signals of different frequency bands can be reconstructed. For example, multi-layer wavelet transform is performed on the original electromyographic signal to decompose the original electromyographic signal into multiple frequency band signals, which can be expressed as:

[0057]

[0058] Among them, x j (t) is the frequency band signal obtained after the j-th wavelet transform is performed on the original electromyographic signal. J represents the number of wavelet transform layers. A J is the approximate coefficient, representing the low-frequency part of the signal, D j is the detail coefficient, which represents the high-frequency component of the signal at the jth layer.

[0059] The preset wavelet coefficient threshold is: Where σ is the standard deviation of the high-frequency detail coefficients and n is the signal length.

[0060] The detail coefficients are denoised based on the preset wavelet coefficient threshold, including: if the detail coefficient is greater than the preset wavelet coefficient threshold, the corresponding detail coefficient is retained; if the detail coefficient is less than or equal to the preset wavelet coefficient threshold, the corresponding detail coefficient is filtered out. After the detail coefficients are thresholded, the approximate coefficients and the denoised detail coefficients are used to perform inverse wavelet transform to obtain the denoised signals of each frequency band:

[0061]

[0062] Where C is a normalization constant.

[0063] The wavelet domain features of the signals in each frequency band after the inverse wavelet transform are extracted, including: extracting wavelet energy features, statistical features such as mean, variance, etc. These features can be used for subsequent muscle movement state analysis.

[0064] It should be noted that adaptive filtering mainly eliminates low-frequency and dynamic noise, while wavelet threshold denoising is more suitable for removing high-frequency noise. The combination of the two can maintain the integrity of the electromyographic signal while filtering out multiple noises. After the embodiment of the present application is processed by adaptive filtering, the time domain features such as sliding window energy and zero-crossing rate are clearer; and after wavelet denoising, the wavelet domain features can more accurately reflect the frequency domain information of the muscle state. The feature vector composed of these features can more comprehensively and accurately describe the changes in the muscle state of the electromyographic signal.

[0065] S104: Fusing the signal energy, zero-crossing rate and wavelet domain features to form a feature vector.

[0066] The above extracted features including signal energy, zero-crossing rate and wavelet domain features are fused to form a feature vector F.

[0067] S105: Input the feature vector into a pre-trained classification model for analysis to obtain the muscle movement state.

[0068] Pre-trained classification models include random forest models. The expressions of the random forest models include:

[0069]

[0070] in, is the preset muscle movement state, RF is the prediction function given the feature vector F and the set of trees T, T is the set of trees in the random forest model, F is the feature vector, T n (F) is the prediction result of the nth tree.

[0071] Through the above analysis, it can be known that the muscle movement analysis method based on electromyographic data provided by the embodiment of the present application includes: using an adaptive filter to pre-process the original electromyographic signal and dynamically filter out the noise signal; using a sliding window to segment the electromyographic signal that has filtered out the noise signal, calculating the energy of the signal in each window and detecting the zero-crossing rate of the signal in each window; performing wavelet threshold denoising on the original electromyographic signal to obtain wavelet domain features; fusing the signal energy, zero-crossing rate and wavelet domain features to form a feature vector; inputting the feature vector into a pre-trained classification model for analysis to obtain the muscle movement state. By analyzing the signal energy, zero-crossing rate and wavelet with feature fusion, real-time and effective analysis of the muscle movement state is achieved, and the accuracy of the muscle movement state analysis under dynamic conditions is improved.

[0072] See also Figure 2 As shown, Figure 2 A schematic diagram of the structure of a muscle movement analysis device based on electromyography data provided in an embodiment of the present application. Figure 2It can be seen that the muscle movement analysis device 20 based on electromyography data provided in the embodiment of the present application includes:

[0073] The first denoising module 201 is used to pre-process the original electromyographic signal using an adaptive filter to dynamically filter out noise signals.

[0074] The processing module 202 is used to segment the electromyographic signal from which the noise signal is filtered out using a sliding window, calculate the energy of the signal in each window, and detect the zero-crossing rate of the signal in each window.

[0075] The second denoising module 203 is used to perform wavelet threshold denoising on the original electromyographic signal to obtain wavelet domain features.

[0076] The construction module 204 is used to fuse the signal energy, zero-crossing rate and wavelet domain features to form a feature vector.

[0077] The analysis module 205 is used to input the feature vector into a pre-trained classification model for analysis to obtain the muscle movement state.

[0078] In one embodiment, the processing module 202 includes a computing unit, which is configured to:

[0079] For any window, the time domain signal value in the window is input into the preset energy calculation formula to calculate the energy of the signal in the window; wherein the preset energy calculation formula is:

[0080]

[0081] Among them, E(i) is the energy of the i-th window, reflecting the strength of muscle activity in the window; N is the length of the sliding window, and x(i+j) is the time domain signal value in the window.

[0082] In one embodiment, the processing module 202 includes a detection unit, which is configured to:

[0083] For any window, the time domain signal value in the window is input into the preset zero-crossing rate calculation formula to calculate the zero-crossing rate of the signal in the window; wherein the preset zero-crossing rate calculation formula is:

[0084]

[0085] Where sgn is the sign function, defined as:

[0086]

[0087] Z(i) represents the number of times the signal crosses zero in the i-th window.

[0088] In one embodiment, the second denoising module 203 includes:

[0089] A denoising unit, used for decomposing the original electromyographic signal into multiple frequency band signals based on wavelet transform, and denoising each frequency band signal based on a preset wavelet coefficient threshold;

[0090] The extraction unit is used to perform inverse wavelet transform on the denoised signals of each frequency band based on the approximate coefficients and the denoised detail coefficients, and extract the wavelet domain features of the signals of each frequency band after the inverse wavelet transform.

[0091] In one embodiment, the denoising unit comprises:

[0092] The decomposition subunit is used to perform multi-layer wavelet transform on the original electromyographic signal, and decompose the original electromyographic signal into multiple frequency band signals, wherein the wavelet coefficients of each frequency band signal include approximate coefficients and detail coefficients;

[0093] The denoising subunit is used to denoise the detail coefficients based on a preset wavelet coefficient threshold.

[0094] In one embodiment, the pre-trained classification model includes: a random forest model.

[0095] In one embodiment, the expression of the random forest model includes:

[0096]

[0097] in, is the preset muscle movement state, RF is the prediction function given the feature vector F and the set of trees T, T is the set of trees in the random forest model, F is the feature vector, T n (F) is the prediction result of the nth tree.

[0098] It should be noted that the specific implementation process of the above modules or units can refer to the specific implementation process of each step in the previous method embodiment, which will not be repeated here.

[0099] See also Figure 3 As shown, Figure 3 A schematic block diagram of a muscle movement analysis device based on electromyography data provided in an embodiment of the present application.

[0100] Exemplarily, the muscle movement analysis device 30 based on electromyography data includes a processing module 301 and a memory 302 .

[0101] Exemplarily, the processing module 301 and the memory 302 are connected via a bus 303 , and the bus 303 is, for example, an I2C (Inter-integrated Circuit) bus.

[0102] Specifically, the processing module 301 may be a micro-controller unit (MCU), a central processing unit (CPU) or a digital signal processing module (DSP).

[0103] Specifically, the memory 302 may be a Flash chip, a read-only memory (ROM) disk, an optical disk, a USB flash drive, or a mobile hard disk.

[0104] The processing module 301 is used to run the computer program stored in the memory 302, and implement the steps of the above-mentioned muscle movement analysis method based on electromyography data when executing the computer program.

[0105] Exemplarily, the processing module 301 is used to run the computer program stored in the memory 302, and implement the following steps when executing the computer program:

[0106] Use adaptive filters to pre-process the original electromyographic signals and dynamically filter out noise signals;

[0107] The electromyographic signal after filtering out the noise signal is segmented using a sliding window, the energy of the signal in each window is calculated, and the zero-crossing rate of the signal in each window is detected;

[0108] Performing wavelet threshold denoising on the original electromyographic signal to obtain wavelet domain features;

[0109] The signal energy, the zero-crossing rate and the wavelet domain features are integrated to form a feature vector;

[0110] The feature vector is input into the pre-trained classification model for analysis to obtain the muscle movement state.

[0111] In one embodiment, calculating the energy of the signal in each window includes:

[0112] For any window, the time domain signal value in the window is input into the preset energy calculation formula to calculate the energy of the signal in the window; wherein the preset energy calculation formula is:

[0113]

[0114] Among them, E(i) is the energy of the i-th window, reflecting the strength of muscle activity in the window; N is the length of the sliding window, and x(i+j) is the time domain signal value in the window.

[0115] In one embodiment, detecting the zero-crossing rate of the signal in each window includes:

[0116] For any window, the time domain signal value in the window is input into the preset zero-crossing rate calculation formula to calculate the zero-crossing rate of the signal in the window; wherein the preset zero-crossing rate calculation formula is:

[0117]

[0118] Where sgn is the sign function, defined as:

[0119]

[0120] Z(i) represents the number of times the signal crosses zero in the i-th window.

[0121] In one embodiment, wavelet threshold denoising is performed on the original electromyographic signal to obtain wavelet domain features, including:

[0122] Decomposing the original electromyographic signal into multiple frequency band signals based on wavelet transform, and denoising each frequency band signal based on a preset wavelet coefficient threshold;

[0123] The denoised signals of each frequency band are subjected to inverse wavelet transform based on the approximate coefficients and the denoised detail coefficients, and the wavelet domain features of the signals of each frequency band after inverse wavelet transform are extracted.

[0124] In one embodiment, the original electromyographic signal is decomposed into multiple frequency band signals based on wavelet transform, and each frequency band signal is denoised based on a preset wavelet coefficient threshold, including:

[0125] Perform multi-layer wavelet transform on the original electromyographic signal to decompose it into multiple frequency band signals. The wavelet coefficients of each frequency band signal include approximate coefficients and detail coefficients.

[0126] The detail coefficients are denoised based on a preset wavelet coefficient threshold.

[0127] In one embodiment, the pre-trained classification model includes: a random forest model.

[0128] In one embodiment, the expression of the random forest model includes:

[0129]

[0130] in, is the preset muscle movement state, RF is the prediction function given the feature vector F and the set of trees T, T is the set of trees in the random forest model, F is the feature vector, T n (F) is the prediction result of the nth tree.

[0131] The specific principles and implementation methods of the muscle movement analysis device based on electromyography data provided in the embodiment of the present application are similar to those of the muscle movement analysis method based on electromyography data in the aforementioned embodiment, and will not be repeated here.

[0132] The embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by the processing module, the processing module implements the following steps:

[0133] Use adaptive filters to pre-process the original electromyographic signals and dynamically filter out noise signals;

[0134] The electromyographic signal after filtering out the noise signal is segmented using a sliding window, the energy of the signal in each window is calculated, and the zero-crossing rate of the signal in each window is detected;

[0135] Performing wavelet threshold denoising on the original electromyographic signal to obtain wavelet domain features;

[0136] The signal energy, the zero-crossing rate and the wavelet domain features are integrated to form a feature vector;

[0137] The feature vector is input into a pre-trained classification model for analysis to obtain the muscle movement state. In one embodiment, the energy of the signal in each window is calculated, including:

[0138] For any window, the time domain signal value in the window is input into the preset energy calculation formula to calculate the energy of the signal in the window; wherein the preset energy calculation formula is:

[0139]

[0140] Among them, E(i) is the energy of the i-th window, reflecting the strength of muscle activity in the window; N is the length of the sliding window, and x(i+j) is the time domain signal value in the window.

[0141] In one embodiment, detecting the zero-crossing rate of the signal in each window includes:

[0142] For any window, the time domain signal value in the window is input into the preset zero-crossing rate calculation formula to calculate the zero-crossing rate of the signal in the window; wherein the preset zero-crossing rate calculation formula is:

[0143]

[0144] Where sgn is the sign function, defined as:

[0145]

[0146] Z(i) represents the number of times the signal crosses zero in the i-th window.

[0147] In one embodiment, wavelet threshold denoising is performed on the original electromyographic signal to obtain wavelet domain features, including:

[0148] Based on wavelet transform, the original electromyographic signal is decomposed into multiple frequency band signals, and each frequency band signal is denoised based on a preset wavelet coefficient threshold;

[0149] The denoised signals of each frequency band are subjected to inverse wavelet transform based on the approximate coefficients and the denoised detail coefficients, and the wavelet domain features of the signals of each frequency band after inverse wavelet transform are extracted.

[0150] In one embodiment, the original electromyographic signal is decomposed into multiple frequency band signals based on wavelet transform, and each frequency band signal is denoised based on a preset wavelet coefficient threshold, including:

[0151] Perform multi-layer wavelet transform on the original electromyographic signal to decompose it into multiple frequency band signals. The wavelet coefficients of each frequency band signal include approximate coefficients and detail coefficients.

[0152] The detail coefficients are denoised based on a preset wavelet coefficient threshold.

[0153] In one embodiment, the pre-trained classification model includes: a random forest model.

[0154] In one embodiment, the expression of the random forest model includes:

[0155]

[0156] in, is the preset muscle movement state, RF is the prediction function given the feature vector F and the set of trees T, T is the set of trees in the random forest model, F is the feature vector, T n (F) is the prediction result of the nth tree.

[0157] Wherein, the computer-readable storage medium can be an internal storage unit of the muscle motion analysis device based on electromyographic data in the aforementioned embodiment, such as a hard disk or memory of the muscle motion analysis device based on electromyographic data. The computer-readable storage medium can also be an external storage device of the muscle motion analysis device based on electromyographic data, such as a plug-in hard disk equipped on the muscle motion analysis device based on electromyographic data, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc.

[0158] It should be understood that the terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application.

[0159] It will also be understood that the term "and / or" used in this application and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0160] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A muscle movement analysis method based on electromyography data, characterized in that: include: Use adaptive filters to pre-process the original electromyographic signals and dynamically filter out noise signals; The electromyographic signal after filtering out the noise signal is segmented using a sliding window, the energy of the signal in each window is calculated, and the zero-crossing rate of the signal in each window is detected; Performing wavelet threshold denoising on the original electromyographic signal to obtain wavelet domain features; The signal energy, the zero-crossing rate and the wavelet domain features are integrated to form a feature vector; The feature vector is input into a pre-trained classification model for analysis to obtain the muscle movement state.

2. The muscle movement analysis method based on electromyographic data according to claim 1, characterized in that: The calculating the energy of the signal in each window comprises: For any window, the time domain signal value in the window is input into the preset energy calculation formula to calculate the energy of the signal in the window; wherein the preset energy calculation formula is: Among them, E(i) is the energy of the i-th window, reflecting the strength of muscle activity in the window; N is the length of the sliding window, and x(i+j) is the time domain signal value in the window.

3. The muscle movement analysis method based on electromyographic data according to claim 1, characterized in that: The detecting the zero-crossing rate of the signal in each window comprises: For any window, the time domain signal value in the window is input into the preset zero-crossing rate calculation formula to calculate the zero-crossing rate of the signal in the window; wherein the preset zero-crossing rate calculation formula is: Where sgn is the sign function, defined as: Z(i) represents the number of times the signal crosses zero in the i-th window.

4. The muscle movement analysis method based on electromyographic data according to claim 1, characterized in that: The wavelet threshold denoising is performed on the original electromyographic signal to obtain wavelet domain features, including: Decomposing the original electromyographic signal into multiple frequency band signals based on wavelet transform, and denoising each frequency band signal based on a preset wavelet coefficient threshold; The denoised signals of each frequency band are subjected to inverse wavelet transform based on the approximate coefficients and the denoised detail coefficients, and the wavelet domain features of the signals of each frequency band after inverse wavelet transform are extracted.

5. The muscle movement analysis method based on electromyographic data according to claim 4, characterized in that: The method of decomposing the original electromyographic signal into multiple frequency band signals based on wavelet transform, and denoising each frequency band signal based on a preset wavelet coefficient threshold, comprises: Performing multi-layer wavelet transform on the original electromyographic signal to decompose the original electromyographic signal into multiple frequency band signals, wherein the wavelet coefficients of each frequency band signal include approximate coefficients and detail coefficients; The detail coefficients are denoised based on a preset wavelet coefficient threshold.

6. The muscle movement analysis method based on electromyographic data according to claim 1, characterized in that: The pre-trained classification model includes: a random forest model.

7. The muscle movement analysis method based on electromyographic data according to claim 6, characterized in that: The expression of the random forest model includes: in, is the preset muscle movement state, RF is the prediction function given the feature vector F and the set of trees T, T is the set of trees in the random forest model, F is the feature vector, T n (F) is the prediction result of the nth tree.

8. A muscle movement analysis device based on electromyography data, characterized in that: include: The first denoising module is used to pre-process the original electromyographic signal using an adaptive filter to dynamically filter out noise signals; A processing module, used for segmenting the electromyographic signal from which the noise signal is filtered out using a sliding window, calculating the energy of the signal in each window and detecting the zero-crossing rate of the signal in each window; A second denoising module is used to perform wavelet threshold denoising on the original electromyographic signal to obtain wavelet domain features; A construction module, used for fusing the signal energy, the zero-crossing rate and the wavelet domain features to form a feature vector; The analysis module is used to input the feature vector into a pre-trained classification model for analysis to obtain the muscle movement state.

9. An electronic device, characterized in that: include: Memory and processing modules; The memory is used to store computer programs; The processing module is used to execute the computer program and implement the steps of the muscle movement analysis method based on electromyography data as described in any one of claims 1 to 8 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer readable storage medium stores a computer program; When the computer program is executed by one or more processing modules, the one or more processing modules execute the steps of the muscle movement analysis method based on electromyography data as described in any one of claims 1 to 8.

Images