The Study Of Acquisition Scneme Of SEMG From Residual Limb And Gait Recognition

With the rapid development of technology and the urgent need of life,the transfemoral amputees are eager to wear intelligent prosthesis.Only if we recognize the different gait accurately can we offer the effective control signal source for the intelligent prosthesis.Based on the above,the prosthesis can realize the function of ‘feedback’ and ‘think’.Surface electromyography(sEMG)signals are used to recognize the different gait as signal source since they can reflect motion intentions better than motion information.In order to recognize the different gait of the transfemoral amputee with the sockets,it is important to record high-quality sEMG from transfemoral amputees with the sockets.In this thesis,a acquisition scheme of sEMG from the residual limb in different gait is designed,then recognize the different gait,including level-ground walking,upstairs,downstairs,upslope and downslope.The research contents are as follows:Firstly,an experimental subject is selected from fifteen single transfemoral amputees investigated,whose residual limb is in good conditions.Residual limb muscle group are selected and the TrignoTM Wireless EMG is determined as the acquisition device.Then,according to the conditions of residual limb,a socket-sensor configuration for the subject is designed.The sEMG signal acquisition system is based on LabView and sEMG signals from the transfemoral amputee with the socket are recorded in different gait.Secondly,after signal preprocessing including filtering and rectifying,a method of Teager-Kaiser energy(TKE)operator is proposed to detect the onset time of muscle activity.The features,including time domain,frequency domain and autoregressive(AR)model parameters,are extracted from the sEMG signals during pre-action period to recognize the different gait,and fuse these features.Then,the different feature vectors are constructed.Finally,a clustering algorithm based on supervised Kohonen neural network is proposed to recognize the different gait.Input the different feature vectors to the network and recognize the different gait.Then,the above recognition rate is contrasted with the recognition rate of Kohonen neural network and the algorithm of K Nearest Neighbor.The experiment result shows the algorithm of S_Kohonen is effective to recognize the different gait,the fusion of features is inevitable and the acquisition scheme of sEMG is feasible.