Research On Recognition Method Of Upper Limb Motor Imagery EEG And Brain-Computer Interface System

Brain-computer interface(BCI)as a new human-computer interaction technology has received great attention in recent decades.It can help patients with severe motor dysfunction communicate with the outside world without relying on peripheral nerves and muscles,and is of great significance for improving the quality of life of patients.How to effectively process EEG signals to achieve accurate recognition is very important to the application of BCI.In this thesis,the upper-limb motor imaging EEG signal is the research object,and the feature extraction of the few-channel motor imaging EEG is studied.In addition,a portable,low-cost brain-computer interface system is designed.The main research contents are as follows:The Power Spectral Density estimation(PSD)is used to extract the frequency domain characteristics of the motor imagery EEG signal,and the Empirical Mode Decomposition(EMD)algorithm is used to remove the components that are not related to the motor imagery EEG.According to different visual cues,the EEG rhythm of the sensory motor area of the brain can be adjusted.An experimental paradigm based on dynamic visual cues is designed and compared with traditional static visual cues.The results show that compared with static visual cues,the EEG signal characteristics of subjects under dynamic visual cues are more separable,which is conducive to the subsequent feature classification.In order to improve the classification performance of motor imagery EEG signals in a few channels,an improved feature extraction method based on Variational Mode Decomposition(VMD)is proposed.The original EEG signal is preprocessed,and the Continuous Wavelet Transform(CWT)is used to analyze the time-frequency characteristics of the EEG signal,and then the EEG signal is decomposed by the VMD algorithm to divide the frequency band of the signal,and calculate the original signal and each signal through the Bhattacharyya distance.The similarity between the Intrinsic Mode Function(IMF)components,the irrelevant IMF components are eliminated,the appropriate IMF components are selected to construct a new signal matrix as the optimal time-frequency combination,and Common Spatial Pattern(CSP)filtering is used to extract the features.and finally Support Vector Machine(SVM)is used to realize classification.The experimental results show that after processing the BCI competition data set,the proposed method can effectively extract more features related to motor imagination,thereby improving the classification accuracy rate,and compared with other algorithms.At the same time,it was verified in the measured data to prove the effectiveness of the method.With portability and low cost as the design goals,the manipulator control system based on the brain-computer interface of motor imagination was designed and realized.The feasibility of the system was verified through online experiments,and effective online EEG processing algorithms and reasonable manipulator control strategies were designed.The average classification and recognition rate of the three subjects was82.8%,which achieved more accurate manipulator control.