CN-122004783-A - Muscle fatigue threshold detection method based on surface electromyographic signals

CN122004783ACN 122004783 ACN122004783 ACN 122004783ACN-122004783-A

Abstract

The invention relates to the technical field of biomedical signal processing, and discloses a muscle fatigue threshold detection method based on surface electromyographic signals, which comprises the steps of acquiring synchronous surface electromyographic signals and angular velocity signals, and dividing to generate a motion cycle sequence; establishing a baseline kinematic envelope template based on initial angular velocity data, carrying out morphological verification on a current angular velocity segment, calculating actual electromechanical delay after the current angular velocity segment passes the verification, comparing the reference delay to obtain time drift quantity, defining an electromyographic signal interception time boundary by combining an initial matching window, intercepting signal segments in the calibration boundary, calculating a frequency band energy distribution index and generating a time sequence, reconstructing the sequence into a two-dimensional phase space state vector, calculating a maximum Lyapunov index, judging that dynamic topological bifurcation occurs when a continuous counting mechanism is met, and outputting a control intervention instruction. The invention can dynamically compensate electromechanical phase dislocation caused by fatigue, and realize accurate detection and safe intervention of muscle fatigue state.

Inventors

Weng Yihan
WANG PING
CHEN YU
YE ZHIHAN
HUANG XIUZHAN

Assignees

阳光学院

Dates

Publication Date: 20260512
Application Date: 20260330

Claims (10)

1. The muscle fatigue threshold detection method based on the surface electromyographic signals is characterized by comprising the following steps of: acquiring a surface electromyographic signal sequence and an angular velocity signal sequence, and dividing to generate a motion cycle sequence; establishing a baseline kinematic envelope template based on angular velocity data of an initial number of motion cycles in the sequence of motion cycles, defining an initial kinematic phase matching window and a reference delay time; Extracting a surface electromyographic signal segment and an angular velocity signal segment of a current motion period, performing morphological verification on the angular velocity signal segment and the baseline kinematic envelope template, and calculating actual electromechanical delay after the morphological verification passes; obtaining electromechanical delay time drift amount by making a difference between the actual electromechanical delay and the reference delay time, and defining an electromyographic signal interception time boundary by combining the initial kinematic phase matching window; Intercepting the surface electromyographic signal fragments within the electromyographic signal interception time boundary to calculate a frequency band energy distribution index, and generating a frequency band energy distribution index time sequence; reconstructing the time sequence of the frequency band energy distribution index into a two-dimensional phase space state vector, calculating the maximum Lyapunov index, and outputting a control intervention instruction when the maximum Lyapunov index meets a continuous counting mechanism.
2. The method for detecting the muscle fatigue threshold based on the surface electromyographic signals according to claim 1, wherein the steps of obtaining the surface electromyographic signal sequence and the angular velocity signal sequence and dividing the same to generate a motion cycle sequence specifically comprise: filtering pretreatment is carried out on the surface electromyographic signal sequence and the angular velocity signal sequence which are synchronously acquired; Traversing data points of the filtered angular velocity signal sequence according to time sequences, extracting the moment when the signal value passes through zero level, and marking the moment as a zero crossing point; Taking two adjacent in-phase zero crossing points on a time axis as start-stop boundaries of a motion interval, reading start zero crossing point time stamps and end zero crossing point time stamps of the motion interval, and respectively intercepting a surface electromyographic signal sequence and an angular velocity signal sequence after filtering in the start-stop boundaries to obtain corresponding surface electromyographic signal segments and angular velocity signal segments; calculating the difference value between the end zero-crossing point time stamp and the start zero-crossing point time stamp, and obtaining the corresponding actual time length; And binding and storing the surface electromyographic signal fragments and the angular velocity signal fragments which contain timestamp information, the corresponding period indexes and the actual time length, so that the filtered signal sequence is cut into the discrete motion period sequence.
3. The method for detecting a muscle fatigue threshold based on surface electromyographic signals according to claim 2, wherein the establishing a baseline kinematic envelope template based on angular velocity data of an initial number of motion cycles in the sequence of motion cycles defines an initial kinematic phase matching window and a reference delay time, specifically comprises: extracting continuous angular velocity signal fragments and surface electromyographic signal fragments in the initial number of motion periods in the motion period sequence as sample data; Performing time axis normalization processing on the angular velocity signal sequences in the initial number of motion periods, and averaging waveform amplitude values of the angular velocity signal fragments in the initial number of motion periods at each relative time node to generate the baseline kinematic envelope template representing a target action force-generating rule; Extracting the peak angular velocity of the baseline kinematic envelope template, searching a continuous data segment with the amplitude between a value obtained by multiplying the peak angular velocity by a set amplitude lower limit proportionality coefficient and a value obtained by multiplying the peak angular velocity by a set amplitude upper limit proportionality coefficient, establishing a section covered by the continuous data segment on a relative period progress axis as the initial kinematic phase matching window, and recording the relative initial time percentage and the relative end time percentage of the initial kinematic phase matching window on a normalized time axis; And respectively extracting the initial points of the surface myoelectric activities of the initial number of movement periods and the mechanical force-generating extreme points of the corresponding angular velocity signal segments, calculating the absolute time difference between the initial points of the surface myoelectric activities and the mechanical force-generating extreme points, calculating an arithmetic average value of the measured absolute time difference, and taking the arithmetic average value as the reference delay time.
4. The method for detecting a muscle fatigue threshold based on a surface electromyographic signal according to claim 3, wherein the morphological verification of the angular velocity signal segment with the baseline kinematic envelope template specifically comprises: resampling the angular velocity waveform of the current period through an interpolation algorithm to enable the data length of the resampled angular velocity waveform of the current period to be the same as the data length of the baseline kinematic envelope template; calculating a pearson correlation coefficient between the resampled angular velocity waveform of the current period and the baseline kinematic envelope template; when the pearson correlation coefficient is smaller than a set isomorphism judging threshold value, reading the stored frequency band energy distribution index of the previous motion period, and directly outputting the frequency band energy distribution index of the previous motion period as the frequency band energy distribution index of the current motion period; and when the pearson correlation coefficient is equal to or larger than the isomorphism judgment threshold value, judging that the current action is valid, and checking the pass.
5. The method for detecting a muscle fatigue threshold based on a surface electromyographic signal according to claim 4, wherein the calculating the actual electromechanical delay after the morphological school passes comprises: carrying out nonlinear transformation on the surface electromyographic signal fragments in the current movement period by adopting a Teager-Kaiser energy operator, obtaining an instantaneous energy sequence, extracting a resting data window in an initial stage, calculating the mean value and standard deviation of background noise in the resting data window, and taking the mean value plus the standard deviation of a set multiple as a self-adaptive trigger threshold; searching data points which continuously exceed the self-adaptive trigger threshold for the first time and the duration reaches the set duration in the instant energy sequence, and recording the time stamp corresponding to the data points as a nerve-excited pacing point time stamp; Carrying out first-order difference derivation on the angular velocity signal of the current period before resampling to obtain an angular acceleration time sequence, extracting a timestamp corresponding to the maximum angular acceleration amplitude point in the angular acceleration time sequence, and recording the timestamp as a mechanical stress extreme point timestamp; and calculating the absolute time difference between the mechanical force extreme point time stamp and the nerve excitation pacing point time stamp, and obtaining the actual electromechanical delay of the current motion cycle.
6. The method for detecting a surface electromyographic signal-based muscle fatigue threshold according to claim 5, wherein the step of obtaining the electromechanical delay time drift by making the difference between the actual electromechanical delay and the reference delay time, and defining an electromyographic signal interception time boundary in combination with the initial kinematic phase matching window, specifically comprises: Calculating the difference value between the actual electromechanical delay of the current motion period and the reference delay time, and obtaining the electromechanical delay time drift amount; Acquiring a starting zero crossing point time stamp and an actual time length of a current motion period, calculating a product of the actual time length and the relative starting time percentage, adding the product with the starting zero crossing point time stamp, and acquiring a starting basic time point; Calculating the product of the actual duration and the relative ending time percentage, and adding the product with the starting zero-crossing point time stamp to obtain an ending basic time point; Subtracting the electromechanical delay time drift amount from the starting basic time point and the ending basic time point respectively to obtain compensated electromyographic signal interception starting time and compensated electromyographic signal interception ending time, thereby defining the electromyographic signal interception time boundary.
7. The method for detecting a muscle fatigue threshold based on a surface electromyographic signal according to claim 6, wherein the step of calculating a frequency band energy distribution index by intercepting the surface electromyographic signal segments within the electromyographic signal interception time boundary, and generating a time sequence of frequency band energy distribution indexes specifically comprises: performing zero filling operation on the intercepted surface electromyographic signal fragments until the total data point number is equal to the analysis point number, wherein the analysis point number is a numerical value which is greater than the maximum fragment length of the signal sequence and is the integral power of 2; Performing discrete Fourier transform on the zero-padded electromyographic signal sequence to obtain a frequency domain complex sequence to calculate power spectrum density, and reserving numerical points corresponding to a frequency interval of 20-500 Hz to generate a power spectrum density sequence; Extracting surface electromyographic signal fragments of the initial number of motion periods, calculating an initial power spectrum median frequency as a critical division frequency, and dividing a frequency interval from 20 Hz to 500 Hz into a low-frequency band interval and a high-frequency band interval according to the critical division frequency; respectively carrying out discrete accumulation summation on the power spectrum density sequences in the low frequency band interval and the high frequency band interval to obtain a low frequency band energy integrated value and a high frequency band energy integrated value of the current motion period; And calculating the ratio of the low-frequency band energy integrated value to the high-frequency band energy integrated value to obtain a frequency band energy distribution index, and arranging the frequency band energy distribution indexes of each continuous motion period in time sequence to generate the frequency band energy distribution index time sequence.
8. The method for detecting a muscle fatigue threshold based on surface electromyographic signals according to claim 7, wherein the reconstructing the frequency band energy distribution index time series into a two-dimensional phase space state vector specifically comprises: Calculating the optimal delay time required by phase space reconstruction of the frequency band energy distribution index time sequence by adopting a mutual information method; selecting a current data point from the frequency band energy distribution index time sequence, and extracting a lag data point which is different from the current data point by the optimal delay time span on a time axis; and the current data point and the lagged data point are combined to form a coordinate component of a two-dimensional vector, and the original one-dimensional scalar data sequence is converted into a two-dimensional phase space state vector which is continuously arranged on a time axis through traversing the time sequence.
9. The method for detecting a muscle fatigue threshold based on surface electromyographic signals according to claim 8, wherein the calculating the maximum lyapunov exponent specifically comprises: traversing the reconstructed two-dimensional phase space state vector, for the selected reference state vector, searching a nearest neighbor state vector which is closest to the reference state vector in space, and defining that the index difference value of the reference state vector and the nearest neighbor state vector on a time axis is larger than a set time separation window parameter; The method comprises the steps of obtaining an evolution step length from zero to a set maximum evolution step length, moving a reference state vector and a nearest neighbor state vector backwards on a time sequence at the same time for the corresponding period number of the evolution step length, calculating Euclidean distance between two new state vectors after evolution movement, and obtaining an evolution distance; Respectively calculating the average value of the natural logarithm of the evolution distance of all the reference state vector pairs, obtaining the average logarithmic dispersion values corresponding to different evolution step sizes, and generating a local dispersion evolution curve with the abscissa as the evolution step size and the ordinate as the average logarithmic dispersion value; and extracting a linear ascending region of the local divergence evolution curve in an initial evolution stage, performing linear fitting on data points of the linear ascending region by adopting a least square method, calculating to obtain a slope of a fitting straight line, and taking the slope as a maximum Lyapunov exponent.
10. The method for detecting a muscle fatigue threshold based on surface electromyographic signals according to claim 9, wherein the outputting a control intervention command when a continuous counting mechanism is satisfied based on the maximum lyapunov exponent specifically comprises: Calculating the maximum Lyapunov indexes of the initial number of motion periods respectively, taking an arithmetic average value as an initial Lyapunov index reference value, multiplying the initial Lyapunov index reference value by a set proportionality coefficient margin, and generating a bifurcation judging threshold value; setting an anomaly counter, and comparing the maximum Lyapunov exponent of the current motion period with the bifurcation judgment threshold value in a numerical mode; If the maximum Lyapunov exponent is greater than or equal to the bifurcation decision threshold, adding one to the anomaly counter value; If the value is smaller than the bifurcation judging threshold value, resetting an abnormality counter; Judging whether the current value of the abnormal counter reaches a set fault-tolerant cycle number, and when the current value reaches the fault-tolerant cycle number, judging that the continuous counting mechanism is met, determining that dynamic topological bifurcation occurs to the target muscle tissue, and further generating the control intervention instruction; And packaging the control intervention instruction into a standard data frame format, sending the control intervention instruction to a receiving terminal device through a communication interface, wherein the receiving terminal device is rehabilitation exoskeleton equipment, front-end graphical user interface equipment or functional electric stimulation equipment, the control intervention instruction sent to the rehabilitation exoskeleton equipment is a resistance adjustment instruction, the control intervention instruction sent to the front-end graphical user interface equipment is a state update data packet, and the control intervention instruction sent to the functional electric stimulation equipment is a shutdown instruction.

Description

Muscle fatigue threshold detection method based on surface electromyographic signals Technical Field The invention relates to the technical field of biomedical signal processing, in particular to a surface electromyographic signal-based muscle fatigue threshold detection method. Background The surface electromyographic signals are used as an objective index capable of noninvasively reflecting the electrophysiological activity of the neuromuscular system and are widely applied to the fields of rehabilitation, exercise biomechanical analysis, exoskeleton robot control and the like. When a human body performs periodic continuous movement, the electromyographic signal segments in each movement period are usually required to be intercepted, and the fatigue state of a target muscle group is estimated by extracting time-frequency domain characteristics or nonlinear kinetic indexes. Existing muscle fatigue detection methods typically rely on signal sampling over a fixed time window set in an initial, non-fatigued state. However, when a human body performs continuous dynamic operation, physiological functions and kinematic characteristics of the human body dynamically change as muscle fatigue is continuously accumulated. On the one hand, fatigue can lead to a decrease in muscle contraction rate, and intervention of neural compensatory mechanisms can deform the motion trajectories, resulting in actual time consumption of a single motion and changes in motor rhythms. On the other hand, accumulation of metabolic products inside the muscle (e.g., lactic acid) reduces the release and recovery efficiency of calcium channels inside the muscle fiber, resulting in an increase in the lag time between the conduction of the nerve excitation command to the generation of the actual mechanical force, i.e., an increase in the electromechanical delay. Because the existing method lacks a dynamic compensation mechanism aiming at the evolution of the time domain parameters, the sampling interval is deviated from the actual electrophysiological force phase caused by continuously intercepting the electromyographic signals along a fixed time boundary. This phase misalignment results in the truncated signal segment not corresponding to the actual contraction phase of the muscle, thus introducing a large amount of extraneous data of non-target phase. If nonlinear phase space reconstruction and Lyapunov index calculation are directly carried out on a signal sequence with phase dislocation, phase space track faults are caused, so that the finally output dynamic characteristics can not reflect the physiological decay rule of muscles, misjudgment or missed judgment on a critical threshold value of muscle fatigue is caused, and timely and safe control intervention is difficult to provide. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a muscle fatigue threshold detection method based on a surface electromyographic signal, which solves the problems that in the prior art, in continuous dynamic movement, the muscle fatigue deepens to cause movement rhythm change and electromechanical delay increase, so that the traditional fixed time window is adopted to intercept electromyographic signals to generate stress phase dislocation, and further the muscle fatigue state assessment is inaccurate. In order to achieve the above purpose, the invention is realized by the following technical scheme: The invention provides a muscle fatigue threshold detection method based on a surface electromyographic signal, which comprises the following steps: acquiring a surface electromyographic signal sequence and an angular velocity signal sequence, and dividing to generate a motion cycle sequence; establishing a baseline kinematic envelope template based on angular velocity data of an initial number of motion cycles in a sequence of motion cycles, defining an initial kinematic phase matching window and a reference delay time; Extracting a surface electromyographic signal segment and an angular velocity signal segment of a current motion period, performing morphological verification on the angular velocity signal segment and a baseline kinematic envelope template, and calculating actual electromechanical delay after the morphological verification passes; The actual electromechanical delay is subjected to time difference with the reference delay to obtain the electromechanical delay time drift amount, and an electromyographic signal interception time boundary is defined by combining an initial kinematic phase matching window; Intercepting surface electromyographic signal fragments within an electromyographic signal interception time boundary to calculate a frequency band energy distribution index, and generating a frequency band energy distribution index time sequence; Reconstructing the time sequence of the frequency band energy distribution index into a two-dimensional phase space state vector, calculating the maximum Lyapunov inde