Fatigue Analysis Of Lower Limb Muscle Based On Surface EMG Signal

Muscle fatigue is a temporary decline in muscle strength during contraction and occurs in nerve or muscle fiber cells.Decreased nutrition and accumulation of metabolites can lead to muscle fiber fatigue,and muscle activity time can also lead to fatigue.To improve the understanding of muscle fatigue is helpful to make a reasonable exercise plan and avoid muscle injury caused by excessive exercise.Many of the current studies are based on the active work to make muscle fatigue before the EMG signal research.But the treadmill this kind of fixed movement load,the passive movement causes the muscle fatigue situation the muscle fatigue research to be relatively scarce.Moreover,there may be differences in the changes of EMG characteristics between active and passive work-induced fatigue.Therefore,the research is carried out from the treadmill exercise experiment,moreover,the EMG feature extraction is different from the traditional method.In this paper,the EMG feature is based on a single gait cycle.The research contents are as follows:1)This paper studies the changes of the electromyography(EMG)characteristics of the lower limb muscles of the subjects during treadmill exercise,and presents the dynamic changes of the EMG characteristics based on gait cycle before and after exercise.Different from the traditional EMG feature extraction,the EMG signal is divided according to the gait cycle,and then the EMG signal in a single gait cycle is taken as the unit,four electromyography features,including Amplitude(EMG),integrated electromyography(IEMG),median frequency(MF)and Sample Entropy,were analyzed.By comparing the differences of EMG characteristics before and after the experiment,we found that the changes of the subjects’ emg characteristics were not very obvious in the walking speed experiment.After the running exercise experiment,the lower limb muscle entered the state of fatigue,the EMG amplitude,the integrated EMG value and the sample entropy had obvious changes,the tibialis anterior muscle(TA)and soleus muscle(SOL)of most of the subjects showed an increase in EMG amplitude and integrated EMG value,especially in IEMG value,the sample entropy decreased after muscle fatigue,but the change of median frequency was not obvious in this experiment.2)The feasibility of four kinds of EMG features based on gait cycle in fatigue assessment and related data classification was studied.The results show that these four EMG characteristics are effective in fatigue assessment and classification of motion velocity.In this study,three algorithms,namely ELM,Kelm and RELM,were used to classify the movement speed and fatigue of the constructed EMG data set,the conclusion that the KELM algorithm has the best classification effect among the three algorithms is obtained when the EMG feature data set is classified.The first step is to construct the EMG characteristic data set at different velocity,and use three neural network algorithms to classify the velocity.The feature data set is composed of four features: EMG amplitude,EMG integral value,median frequency and sample entropy.It is found that Kelm has the best effect in the classification of motion velocity,the KELM classification accuracy was above 90% for most of the subjects’ datasets.On this basis,the EMG characteristic data sets at the beginning and the end of the exercise experiment are constructed,and the fatigue classification is made according to the EMG characteristic data before and after the exercise experiment.Based on the comparison of the classification accuracy of the data,we found that the Kelm algorithm has the best classification performance in the fatigue classification experiment,and the average classification accuracy of most subjects’ data sets is over 70%.3)A gait EMG data analysis system based on MATLAB GUI is developed,e MG and gait position signal preprocessing module,Signal Division module,feature extraction module and data statistical analysis module are integrated in a software platform.This system provides a basis for the preliminary processing of EMG signals and simple statistical analysis of EMG characteristics,and can provide some help for researchers related to EMG signals.