Wearable Human Activity Recognition System Based On SEMG Signal

Human activity recognition(HAR)technology can fully reflect the human movement and physiological function,which is a great of significance to individual behavior research.In recent years,wearable sensor-based human activity recognition technology has attracted the widespread attention.Compared with the traditional technology,the wearable sensor-based human activity recognition technology has the advantages of good mobility,long life time and strong anti-interference ability,and has been widely used in medical rehabilitation,sports tracking,smart home,human-computer interaction and so on.However,the current sensors used for human activity recognition are mostly based on indirect signals generated by human movements,such as acceleration,electrocardiogram and posture.In recent years,electromyography(EMG)is widely used in the field of prosthetic control and muscle fatigue analysis.EMG is a bioelectric signal generated when muscles contract,which is a direct reflection of the human activities.The information of EMG contained is a represent of the human movements.Therefore,this thesis centers on human activity recognition using surface EMG(sEMG)signal.In order to achieve long-term and real-time continuous acquisition of sEMG signal,this thesis first designs a wearable signal acquisition system for collecting sEMG signal.The wearable signal acquisition system is small,light-weight and low power consumption,to meet the daily human sEMG signal acquisition tasks.SubseqLuently,the thesis uses the wearable signal acquisition system to classify the four kinds of the human activities of walking,running,upstairs and downstairs,and the average recognition accuracy of classification is as high as 98.55%.The experimental results show that the wearable human activity recognition system based on sEMG signal can work normally,and the feature extraction method and pattern recognition strategy are effective.The research done in this thesis provides references for the wearable human activity recognition technology.