阅读说明：本技术 一种肌肉运动单元搜索方法 (A kind of muscular movement unit searches method ) 是由 何金保 安鹏 胡庆波 骆再飞 于 2019-09-11 设计创作，主要内容包括：本发明提供一种肌肉运动单元搜索方法,首先采集多通道表面肌电信号,然后分离波形传输分量和非传输分量,最后利用非传输分量相关性计算发放序列,搜索肌肉运动单元。由于肌肉运动单元发放波形中非传输分量一致性好,从非传输分量角度搜索肌肉运动单元可靠性高,并且本发明提供的肌肉运动单元搜索方法可以应用于肌肉动态收缩的情况。同时,在传输分量与非传输分量分离过程中,采用遗传算法保正结果最优,提高了肌肉运动单元搜索准确性。本发明实现简单,满足实际应用的需要。(The present invention provides a kind of muscular movement unit searches method, first acquisition multi-channel surface myoelectric signal, then separates waveform transfer component and non-transmitting component, is finally calculated using non-transmitting component correlations and provide sequence, searches for muscular movement unit.Since to provide non-transmitting component consistency in waveform good for muscular movement unit, from non-transmitting component angle searching muscular movement unit high reliablity, and the case where muscular movement unit searches method provided by the invention can be applied to muscle Dynamic contraction.It is optimal using the positive result of genetic algorithm guarantor meanwhile in transmission component and non-transmitting component separation procedure, improve muscular movement unit searches accuracy.The present invention realizes simply, meets the needs of practical application.)

1. a kind of muscular movement unit searches method, it is characterized in that the following steps are included:

Step 1 acquires the multi-channel surface myoelectric signal under contraction of muscle, filters to signal, weakens interference；

Amplitude thresholds are arranged in step 2, extract each channel and provide waveform；

Step 3 calculates the inverse matrix of signal cross-correlation matrix for surface electromyogram signal；

Step 4, using surface electromyogram signal residual error and non-transmitting signal energy as optimization object function, wave is provided in separation Shape transmits component and non-transmitting component, objective function are as follows:

Wherein y, K, W, X, λ, X₀Respectively surface electromyogram signal, weighting coefficient matrix, intensity, signal array, regularization parameter and Non-transmitting component；

Step 5 obtains providing sequence according to signal correlation, it is as follows to provide sequence ξ (n) calculation formula:

Wherein X₀It (n) is the non-transmitting component of muscular movement unit, T indicates transposition, C^-1It is the inverse matrix of signal cross-correlation matrix, y It is surface electromyogram signal, n is that signal corresponds to the moment；

Step 6 rejects repetition and unreasonable muscular movement unit, optimization knot finally to all moving cell classified finishings Fruit.

2. a kind of muscular movement unit searches method according to claim 1, which is characterized in that the objective function of step 4 Middle regularization parameter λ is obtained by S curve, since the λ value obtained from S curve is a range, in operation, certain In the range of using genetic algorithm search λ optimal value.

Technical field

The present invention relates to a kind of muscular movement unit searches methods.

Background technique

Surface electromyogram signal (surface EMG, sEMG) is to detect electromyography signal, flesh from body surface using surface electrode The granting of meat moving cell is the basis for constituting sEMG, and obtaining moving cell details by sEMG helps to understand Neuromuscular Meat working mechanism.The searching method of current muscular movement unit (Motor Unit, MU) is sEMG Signal averaging is serious, interference By force, when MU granting frequency is close, the muscular movement element characteristics parameter of acquisition is undesirable.Especially in muscle Dynamic contraction, flesh The search of meat moving cell is more difficult.How when more MU is provided, improves acquisition MU parameters precision and need further to be studied.

The multichannel sEMG signal processing method of domestic and foreign scholars' research mainly has: K mean cluster algorithm, template matching Method, artificial neural network (ANN) algorithm, real time linear aliasing blind signal separation algorithm, independent element point folding (ICA), convolution kernel The methods of backoff algorithm.As a whole, using multi-channel surface myoelectric signal search muscular movement unit also in the exploratory stage, It is one of the difficult point of myoelectricity research field.

Summary of the invention

In view of the above problems, the purpose of the present invention is to provide a kind of muscular movement unit searches methods, it is characterized in that packet Include following steps:

Step 1 acquires the multi-channel surface myoelectric signal under contraction of muscle, filters to signal, weakens interference；

Amplitude thresholds are arranged in step 2, extract each channel and provide waveform；

Step 3 calculates the inverse matrix of signal cross-correlation matrix for surface electromyogram signal；

Step 4, using surface electromyogram signal residual error and non-transmitting signal energy as optimization object function, separation is sent out Waveform transfer component and non-transmitting component are put, objective function is as follows:

Wherein y, K, W, X, λ, X₀Respectively surface electromyogram signal, weighting coefficient matrix, intensity, signal array, regularization Parameter and non-transmitting component；

Step 5 obtains providing sequence according to signal correlation, it is as follows to provide sequence ξ (n) calculation formula:

Wherein X₀It (n) is the non-transmitting component of muscular movement unit, T indicates transposition, C^-1It is the inverse square of signal cross-correlation matrix Battle array, y is surface electromyogram signal, and n is that signal corresponds to the moment；

Step 6 rejects repetition and unreasonable muscular movement unit finally to all moving cell classified finishings, excellent Change result.

The measure of optimization includes:

Regularization parameter λ is obtained by S curve in the objective function of step 4, since the λ value obtained from S curve is one Range searches λ optimal value using genetic algorithm in a certain range in operation.

Since muscular movement unit is provided in waveform, non-transmitting component is for transmission component, and consistency is more preferable, i.e., Make in the case of being muscle Dynamic contraction, non-transmitting component stability is also fine, and therefore, the present invention provides a kind of based on non-transmitting point The muscular movement unit searches method of amount.In transmission component and non-transmitting component separation procedure, positive knot is protected using genetic algorithm Fruit is optimal, improves muscular movement unit searches accuracy.

Detailed description of the invention

Fig. 1 is flow chart of the present invention.

Fig. 2 is the muscular force variation slope fit schematic diagram of the embodiment of the present invention.

Specific embodiment

Below in conjunction with attached Example, present invention is further described in detail, and those skilled in the art can be by this specification institute The content of exposure is realized easily.

It is as shown in Figure 1 flow chart of the invention.The present invention provides a kind of muscular movement unit searches method, it is characterized in that The following steps are included:

Step 1 acquires the multi-channel surface myoelectric signal under muscle Dynamic contraction, filters to signal, weakens interference.By It include various interference signals in sEMG signal, pretreatment retains 10Hz--500Hz frequency range firstly the need of bandpass filter is used Then signal uses notch filter, filters out 50Hz Hz noise.

Amplitude thresholds are arranged in step 2, extract each channel and provide waveform.Amplitude thresholds are arranged as the case may be, high Granting waveform is then extracted in threshold value, is not otherwise extracted, waveform length takes 40ms before and after the corresponding moment.

Step 3 calculates the inverse matrix needle of signal cross-correlation matrix for surface electromyogram signal.Multichannel letter is calculated first Number cross-correlation matrix and cross-correlation matrix inverse matrix, cross-correlation matrix indicate are as follows:

C=E (S (n) S^T(n))

Wherein n is sampling instant, and S (n) is the multi channel signals of n-th of sampling instant, S^TIt (n) is n-th of sampling instant Multi channel signals transposition, E () be number sequence expectation, calculate the inverse matrix C of cross-correlation matrix^-1, i.e.,

C-¹=[E (S (n) S^T(n))]^-1

Step 4, using surface electromyogram signal residual error and non-transmitting signal energy as optimization object function, separation is sent out Waveform transfer component and non-transmitting component are put, objective function is as follows:

Wherein y, K, W, X, λ, X₀Respectively surface electromyogram signal, weighting coefficient matrix, intensity, signal array, regularization Parameter and non-transmitting component.Weighting coefficient matrix K takes 1 in waveform maximum amplitude, and minimum amplitude takes 0.1, in maximum and minimum Then according to being uniformly distributed between amplitude.Intensity W takes 1, and signal array X transmits component, and regularization parameter λ is obtained by S curve, Since the λ value obtained from S curve is a range, in operation, λ is searched most using genetic algorithm in a certain range The figure of merit.As shown in Fig. 2, providing is non-transmitting component in waveform dotted line frame, remainder is transmission component.

Moving cell waveform is made of transmission component and non-transmitting component, and non-transmitting component is disappeared the stage in waveform It generates, according to the waveform similarity of muscle fibre vertical direction top electrode signal as, waveform electrode can be write to transmission component and non- Transmit the weighted superposition form of component.

Step 5 obtains providing sequence according to signal correlation, it is as follows to provide sequence ξ (n) calculation formula:

Wherein X₀It (n) is the non-transmitting component of muscular movement unit, T indicates transposition, C^-1It is the inverse square of signal cross-correlation matrix Battle array, y is surface electromyogram signal, and n is that signal corresponds to the moment.

Step 6 rejects repetition and unreasonable muscular movement unit finally to all moving cell classified finishings, excellent Change result.It calculates and provides sequence, finally to all granting sequence classified finishings, reject repetition and unreasonable granting moment, At the time of less than 15 ms interval, need to reject.

In conclusion the present invention provides a kind of muscular movement unit searches method, since muscular movement unit provides waveform In non-transmitting component relative to transmission component for, consistency is more preferable, even in the case of muscle Dynamic contraction, non-transmitting component Stability is also fine, and therefore, muscular movement unit searches method provided by the invention can be applied to the feelings of muscle Dynamic contraction Condition.It is optimal using the positive result of genetic algorithm guarantor meanwhile in transmission component and non-transmitting component separation procedure, improve muscle Moving cell searching accuracy.The present invention realizes simply, meets the needs of practical application.

The above-described embodiments merely illustrate the principles and effects of the present invention, and is not intended to limit the present invention.It is any ripe The personage for knowing this technology all without departing from the spirit and scope of the present invention, carries out modifications and changes to above-described embodiment.Cause This, institute is complete without departing from the spirit and technical ideas disclosed in the present invention by those of ordinary skill in the art such as At all equivalent modifications or change, should be covered by the claims of the present invention.