CN105147251A - Muscle fatigue dynamic prediction method based on multi-channel sEMG - Google Patents

Abstract

The invention provides a dynamic prediction method of muscle fatigue based on sEMG. Since the sEMG signal is susceptible to external interference, the average value of all channels is first reselected as the reference value, and the new reference value is differentiated from the original multi-channel sEMG, which effectively reduces the external interference. The time-sharing least squares method achieves dynamic update and predicts the time of muscle fatigue through time reference translation. The MPF and MF parameters are important parameters reflecting muscle fatigue. In order to improve the reliability of muscle fatigue prediction, the present invention uses MPF and MF parameters to predict muscle fatigue time, which effectively avoids the instability of individual parameter prediction and improves the prediction accuracy. robustness. The invention has high prediction accuracy, fast operation speed, simple realization and important application value.

Description

Dynamic prediction method of muscle fatigue based on multi-channel sEMG

technical field

The invention relates to a multi-channel sEMG-based dynamic prediction method for muscle fatigue.

Background technique

Research on the mechanism and prediction of neuromuscular fatigue is a hot spot in sports medicine research at home and abroad, and it is also the focus of sports human science research. During exercise, due to the oxygen supply in the blood or the lack of nutrients, a series of changes will occur in the structure, metabolism and energy of the muscle, which will reduce the efficiency of the neuromuscular system, so that the muscle cannot continue to complete the task, resulting in Muscle fatigue. Muscle fatigue may lead to muscle damage, and muscle fatigue in severe cases will not be recoverable. The research on muscle fatigue has broad application prospects in the fields of ergonomics, human-machine interface, rehabilitation medicine, sports injuries, and artificial limbs.

At present, the clinical detection tools for muscle fatigue mainly include myoelectric signal (sEMG, surface electromyography), muscle sound (MMG, Mechanomyogram), myosonography (SMG, Sonomyography), near-infrared spectroscopy (NIRS, Near-infraredspectroscopy), sonogram (AMG, Acousticmyography), angle sensor, etc. Using sEMG to record and study muscle fatigue is a commonly used method in labor physiology. As a simple, non-invasive and quantitative research method, it can study the changing characteristics of local muscle fatigue and is an accurate detection tool. MMG records muscle vibration signals from the skin surface, and it can capture muscle activity well. However, MMG cannot be used for the study of muscle dynamic contraction, and its stability is not high in repeated detection of muscle fatigue. SMG uses ultrasound to describe the changes in the shape and structure of skeletal muscle. This method is suitable as an auxiliary means of sEMG to provide more muscle fatigue information. NIRS uses the near-infrared portion of the electromagnetic spectrum to measure the absorption property of blood, hemoglobin, and the NIRS signal is unreliable due to inconsistent muscle oxygen content during isometric contractions. AMG is a special application of MMG, which records the sound signal of muscle contraction. This technology is not widely used at present and is somewhat outdated. Angle sensor, the angle sensor has short service life and high cost. There are also dynamometers, Moore-Garg indices, and more. Overall, sEMG and MMG signals are the main means of clinical research on muscle fatigue, and sEMG is considered the most suitable.

Since the sEMG signal of muscle fatigue is affected by the following factors: exercise mode, exercise intensity, muscle contraction mode (centrifugal, concentric contraction), muscle type, individual characteristics, etc., in many researches on fatigue EMG characteristics, different characteristics Parameters are not fully consistent, and muscle fatigue prediction remains difficult.

Contents of the invention

In view of the shortcomings of the prior art described above, the purpose of the present invention is to provide a fast method for online prediction of muscle fatigue based on sEMG.

A method for dynamic prediction of muscle fatigue based on multi-channel sEMG, comprising the steps of:

1): Preprocess the multi-channel sEMG signal: reselect the reference value, and use a band-pass filter and a band-stop filter to eliminate interference.

2): The average power frequency MPF and median frequency MF curves of sEMG are fitted by the time-sharing least squares method;

3): Predict muscle fatigue time: according to the MPF and MF fitting results, calculate the muscle fatigue time by weighting the obtained time;

4): Repeat steps 1), 2), and 3) to dynamically update and predict muscle fatigue time.

Optimization measures also include:

Step 1) Preprocessing the multi-channel sEMG signal, the specific steps are as follows:

a) will Channel sEMG signal Add and average ;

b) Using the above average as a reference value, a new channel sEMG signal, , ,..., ;

c) replace the new Channel sEMG signal Perform band-pass filtering and band-stop filtering to eliminate interference, and subsequent sEMG signal processing is aimed at the new Channel sEMG signal.

The band-pass filter is used to retain the 10Hz--500Hz frequency band signal, and then the notch filter is used to filter out the 50Hz power frequency interference. Both the band-pass filter and the band-stop filter adopt conventional methods, using Butterworth digital filters, and setting parameters include order, 3dB cut-off frequency, filter pass band, stop band cut-off frequency, etc.

Step 2) Time-sharing least squares method, the specific method is: firstly take the current time as the benchmark, take the sEMG signal with a duration of K seconds (K is a positive integer) forward, and then calculate the 1st second and the 2nd second for this segment of sEMG , the 3rd second, ..., the average power frequency MPF and the median frequency MF of the K second, and then use the least square method to fit the MPF and MF curves respectively, and finally get the average power frequency MPF on the curve to drop to 90%MPF corresponding moment , and the moment when the median frequency drops to 90% MPF . The parameters of the least squares method can be calculated by the following formula:

in means MPF or MF, Denotes the mean value of MPF or MF, indicates the moment, , represents the time mean.

The calculation formulas of MPF and MF are as follows:

in is the frequency of the electromyographic signal For its power density spectrum, the classical power spectrum estimation technique based on Fourier analysis is used to estimate .

Step 3) Predict muscle fatigue time. The change of the power spectrum of the EMG signal can reflect the change of conduction velocity, so it can also reflect the degree of fatigue, and both MPF and MF have a downward trend with muscle fatigue. The median frequency MF can reflect the conduction velocity more reliably than the average power frequency MPF. MF is less sensitive to noise and is suitable for online estimation. However, MPF is a more stable measurement for spectrum shift, especially if the signal-to-noise ratio is high, MPF is more reliable. In order to improve the reliability of muscle fatigue prediction, the present invention uses two parameters, MPF and MF, to predict muscle fatigue time. According to the above time , , the predicted muscle fatigue time function is:

in is the weighting coefficient, which is selected according to the characteristics of the individual such as age and gender,

Step 4) Repeat steps 1), 2), and 3). Through the translation of the time reference, sEMG of different time periods can be obtained, time-sharing signal analysis, dynamic update, and prediction of muscle fatigue time.

Using the above method to predict muscle fatigue, the prediction is accurate and fast. First, the average value of all channels is reselected as the reference value. Since most of the external interference will affect all electrodes, the difference between the new reference value and the original multi-channel sEMG effectively reduces the external interference, and the reference value changes , will not affect the subsequent prediction operation, but improve the accuracy of the prediction. The adoption of the time-sharing least squares method simplifies the calculation, improves the operation speed, and is simple to implement. The application of MPF and MF parameters effectively avoids the instability of individual parameter prediction and improves the robustness of prediction. Time-sharing signal analysis, dynamic update, and online prediction of muscle fatigue time.

Description of drawings

Fig. 1 is the flow chart of the present invention.

Detailed ways

Embodiments of the present invention are described below through specific examples, and those skilled in the art can easily realize the contents disclosed in this specification.

As shown in Figure 1, a kind of muscle fatigue dynamic prediction method based on multi-channel sEMG, it comprises the following steps:

1): Preprocess the multi-channel sEMG signal: reselect the reference value, and use a band-pass filter and a band-stop filter to eliminate interference. The specific steps are as follows:

a) will Channel sEMG signal Add and average ;

b) Using the above average as a reference value, a new channel sEMG signal, , ,..., ;

c) replace the new Channel sEMG signal Perform band-pass filtering and band-stop filtering to eliminate interference, and subsequent sEMG signal processing is aimed at the new Channel sEMG signal.

The band-pass filter is used to retain the 10Hz--500Hz frequency band signal, and then the notch filter is used to filter out the 50Hz power frequency interference. Both the band-pass filter and the band-stop filter adopt conventional methods, using Butterworth digital filters, and setting parameters include order, 3dB cut-off frequency, filter pass band, stop band cut-off frequency, etc.

2): The average power frequency MPF and median frequency MF curves of sEMG were fitted by the time-sharing least squares method. The specific method is: firstly take the current time as the benchmark, take the sEMG signal with a duration of K seconds (K is a positive integer) forward, and then calculate the 1st second, the 2nd second, the 3rd second, ... for this segment of sEMG, The average power frequency MPF and median frequency MF of the K second, and then use the least square method to fit the MPF and MF curves respectively, and finally obtain the corresponding time when the average power frequency MPF on the curve drops to 90% MPF , and the moment when the median frequency drops to 90% MPF . The parameters of the least squares method can be calculated by the following formula:

in means MPF or MF, Denotes the mean value of MPF or MF, indicates the moment, , time mean.

The calculation formulas of MPF and MF are as follows:

in is the frequency of the electromyographic signal For its power density spectrum, the classical power spectrum estimation technique based on Fourier analysis is used to estimate .

3): Predict the muscle fatigue time: according to the MPF and MF fitting results, calculate the muscle fatigue time by weighting the obtained time. The change of the power spectrum of the EMG signal can reflect the change of conduction velocity, so it can also reflect the degree of fatigue, and both MPF and MF have a downward trend with muscle fatigue. The median frequency MF can reflect the conduction velocity more reliably than the average power frequency MPF. MF is less sensitive to noise and is suitable for online estimation. However, MPF is a more stable measurement for spectrum shift, especially if the signal-to-noise ratio is high, MPF is more reliable. In order to improve the reliability of muscle fatigue prediction, the present invention uses two parameters, MPF and MF, to predict muscle fatigue time. According to the above time , , the predicted muscle fatigue time function is:

in is the weighting coefficient, which is selected according to the characteristics of the individual such as age and gender,

4): Repeat steps 1), 2), and 3) to dynamically update and predict muscle fatigue time. Through the translation of the time base, sEMG of different time periods can be taken, and the time-sharing signal analysis can realize dynamic update and predict the time of muscle fatigue.

In summary, the present invention proposes that the muscle fatigue prediction method based on sEMG can accurately predict the muscle fatigue time. Since the sEMG signal is susceptible to external interference, the multi-channel signal average is used as a new reference value to effectively suppress the interference of the signal. The MPF and MF parameters are important parameters reflecting muscle fatigue. In order to improve the reliability of muscle fatigue prediction, the present invention uses MPF and MF parameters to predict muscle fatigue time. The time-sharing least squares method achieves dynamic update and predicts the time of muscle fatigue through time reference translation. The invention effectively overcomes various shortcomings in the prior art and has important application value.

The above-mentioned embodiments only illustrate the principles and effects of the present invention, but are not intended to limit the present invention. Anyone skilled in the art can modify or change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by those skilled in the art without departing from the spirit and technical ideas disclosed in the present invention should still be covered by the claims of the present invention.

Claims (5)

1., based on a muscle fatigue dynamic prediction method of multichannel sEMG, it is characterized in that, it comprises the steps:

1): pretreatment is carried out to multichannel sEMG signal: reselect reference value, and adopt band filter, band elimination filter, interference is eliminated;

2): the frequency of average power MPF, the median frequency MF curve that adopt timesharing least square fitting sEMG;

3): the prediction muscle fatigue time: according to MPF, MF fitting result, the time weight obtained is calculated the muscle fatigue time;

4): step 1), 2 repeatedly), 3), dynamically update prediction the muscle fatigue time.

2. the muscle fatigue dynamic prediction method based on multichannel sEMG according to right 1, is characterized in that, step 1) carries out pretreatment to multichannel sEMG signal, and concrete steps are as follows:

A) will passage sEMG signal is added, and is averaging ;

B) using above-mentioned meansigma methods as with reference to value, obtain new passage sEMG signal, , ..., ;

C) by new passage sEMG signal carry out bandpass filtering and bandreject filtering, eliminate interference, follow-up sEMG signal processing is all for new passage sEMG signal.

3. the muscle fatigue dynamic prediction method based on multichannel sEMG according to right 1, it is characterized in that, step 2) timesharing method of least square, concrete grammar is: be first benchmark with current time, get forward the sEMG signal of K second (K is positive integer) duration, secondly for this section of sEMG, calculate the 1st second, the 2nd second, the 3rd second ..., the frequency of average power MPF of K second and median frequency MF, then utilize method of least square matching MPF and MF curve respectively, finally obtain frequency of average power MPF on curve respectively and drop to the moment corresponding to 90%MPF , and median frequency drops to the moment corresponding to 90%MPF .

4. the muscle fatigue dynamic prediction method based on multichannel sEMG according to right 1, is characterized in that, the step 3) prediction muscle fatigue time, according to the above-mentioned time , , prediction muscle fatigue time function is:

Wherein be weight coefficient, the feature such as age, sex according to individuality is specifically selected, .

5. the muscle fatigue dynamic prediction method based on multichannel sEMG according to right 1, it is characterized in that, step 4) is step 1), 2 repeatedly), 3), by the translation of time reference, the sEMG of desirable different time sections, time signal analysis, realization dynamically updates, the prediction muscle fatigue time.