Movement Pattern Classification of Surface Electromyography Based on Wavelet Transform and PSO-SVM

In order to improve the accuracy of surface electromyogram movement pattern classification, a new classification method based on the combination of wavelet transform and particle swarm optimization-support vector machine (PSO-SVM) was proposed. Firstly, two surface electromyography signals from channels of brachioradialis muscle and flexor carpi ulnaris were acquired with virtual instruments. Secondly, the wavelet transform was used to decompose the surface electromyography, and the maximum value of wavelet coefficients was extracted as the feature vector of the surface electromyogram movement pattern. Finally, take the features as the input, SVM classifier was employed to classify the surface electromyogram pattern, and in which PSO algorithm was used to optimize the penalty parameter and kernel function of SVM. Experimental results show that four movement patterns of wrist down, wrist up, hand grasps, hand extension are successfully classified with the proposed pattern classification method, which has higher classification accuracy than that of traditional one.