Analysis And Processing Of Motor Imagery EEG In Brain Computer Interface

Brain-Computer Interface(BCI)is a human-computer interaction way based on electroencephalogram(EEG)which can rely on the EEG signals,not the peripheral muscles and nerves tissue to achieve free action and communication with the outside world.In this paper,the research is focused on the motor imagery EEG signals,the aspects of preprocessing,feature extraction and pattern classification on motor imagery EEG signals are studied.The main research work of the paper has been arranged as follows:(1)Preprocessing: In order to reduce the interference of noise and other unexpected components,adaptive wavelet threshold de-noising method has been adopted to eliminate the noise signal,the data from BCI competition has been used to test the effect,and the results show that the adaptive wavelet threshold de-noising method can effectively reduce the noise interference.(2)Feature extraction: a method combining improved EMD with fuzzy entropy has been proposed for motor imagery EEG feature extraction due to the characteristics of nonlinear EEG signals.The original signal is decomposed by MEMD,and the IMF components of the effective information are selected by mutual information,then,the original signal which is reconstructed can be extracted by the algorithm of fuzzy entropy.At last the data from 2008 BCI competition has been used to verify the feature extraction effect.(3)Pattern classification: when using support vector machine for pattern classification,the kernel parameter g and the penalty parameter C of the support vector machine has a close relationship with the performance of the classifier.So,artificial bee colony(ABC)algorithm has been adopted to optimize the model parameters to improve the performance of the classifier.Then,the optimized classifier has been used for classification of motor imagery EEG signals.(4)Experimental results and analysis: in this part,the competition data from 2005 and 2008 BCI Competition have been used to conduct experiment about motor imagery EEG feature extraction and pattern recognition.The first experiment has been carried out to compare the optimization algorithm,which verifies the good performance of the artificial bee colony algorithm.In experiment two,the final results show that the ABC optimized SVM classifier can improve the classification accuracy of EEG signals effectively.In experiment three,the improved EMD combined with fuzzy entropy method has been used for EEG feature extraction,and the ABC optimized SVM classifier has been adopted for pattern recognition.The final results demonstratethat the improved EMD combined with fuzzy entropy algorithm can effectively extract the EEG signals feature,and the classification rate of ABC optimized SVM classifier is higher than the traditional SVM.