Blind Signal Separation Of Muscle Fatigue Contraction

Using dynamic sEMG(Surface electromyographic signal) to evaluate muscle fatigue condition is very important for clinical medicine. sEMG is a mixture signal including the dynamic fatigue contraction signal and the static fatigue contraction. Under the dynamic environment, the contraction of muscle fibers belong to static fatigue, the static fatigue is a smooth process, the contraction of others belong to dynamic fatigue, the dynamic contraction is a nonstationary process. To accurate evaluation of muscle fatigue, it is necessary to separate the static signal and the dynamic signal from sEMG.Adjust the way to muscle contraction, ignore static signal, and then use relevant spectrum estimation technique to evaluate muscle fatigue condition in dynamic environment, but it will lead to many error. BSS(Blind Source Separation) can do it. Based on the mathematical model of fatigue condition,to separate mixed sEMG using NGA(Natural Gradient algorithm) and infromax(information maximization algorithm).The main research as follows:(1) Based on the two kinds of different features muscle contraction signal mixed model. Use three noise sources to produce fatigue signals, and then compare EMG;(2) When muscle fatigue signals the same noise signal, but coefficient is different, use BSS algorithms to separate sEMG, Analyze and compare the result;(3) When coefficient is same, compare separation result use two BSS algorithm.from different noise sources;(4) When coefficient and sEMG are same, compare separation performance with different BSS;Conclusions as follows:(1) The dynamic signal from gauss noise has nonstationary process, it is a most suitable dynamic signal;(2) When algorithm is same, coefficient is different, muscle fatigue signals from different noise has not obvious effect on separation performance;(3) As muscle signals from same noise, regardless of coefficient, the performance is not obvious effective using NGA. Coefficient respectively0.2,0.4,0.6,0.8,1.0. When it is0.2, infromax has best separation performance;(4) When coefficient is same, fatigue signals come from same noise sources, infromax has better performance.