Research on EEG Signal Recognition Method Based on Whale Algorithm Optimized Support Vector Machine

Abstract

Aiming at the low recognition and classification of motor imagery EEG signals, a pattern recognition method based on support vector machine (SVM) optimized by whale algorithm is proposed. Firstly, band-pass filter is used to preprocess the original signal. Secondly, the eigenvectors are extracted by empirical mode decomposition (EMD) - common space pattern (CSP). Finally, the kernel function parameters and penalty factors of SVM were optimized, and the motor imagery EEG signal recognition model was established to classify the grasping, elbow bending and wrist bending actions. Compared with other classification methods such as grid search optimization SVM, k-nearest neighbor and BP neural network, the results show that the recognition accuracy of the proposed method is higher, which proves that the algorithm can effectively visualize the EEG features of motor imagery.