The Research Of Motor Imagery Recognition Combined With Proprioceptive Feedback

Electroencephalogram(EEG)is a kind of biological signal to reflect thinking conscious and physical activity.One important EEG applications research direction is Brain Computer Interface(BCI),which can achieve between the human brain and other peripheral device.BCI Based on motor imagery EEG is more suitable for the rehabilitation aids control and functional rehabilitation aids,due to simple imagine tasks,the high consistency of imagination and target action,and realization of asynchronous communication.It has become a hot research topic in the field of BCI has attracted much attention in recent years.In the BCI system,the most core part is converting EEG to the control command,i.e.the feature extraction and pattern classification.Based on proprioceptive feedback research,the different patterns of proprioceptive feedback are introduced into the process of the recognition,and then analysis the impact of real-time identification and accuracy.Below are the main contents and innovations of this paper:(1)Empirical mode decomposition(EMD)usually occurs the phenomenon of mode mixing,and traditional lifting wavelet easily leads to serious distortion.In order to solve these problems in the noise elimination,this paper presents a de-noising method based on ensemble empirical mode decomposition(EEMD)and improved lifting wavelet.The effectiveness of the proposed method is demonstrated by means of simulation and comparison experiments,laying the foundation for further processing.(2)On account of the limitation of the EMD in the processing of multidimensional signal,a feature extraction method is proposed based on noise-assisted multivariate empirical mode decomposition(NA-MEMD)and sample entropy.Considering the rhythm waves corresponding to the event-related desynchronization(ERD)/ event-related synchronization(ERS)phenomenon,extract sample entropy of C3 and C4 channels as vector.(3)Extreme Learning Machine(ELM),is applied in the classification after feature component extraction.ELM is an algorithm to solve single hidden layer neural network.It is studied that how to determine the excitation function and the number of hidden layer neurons.Finally,the experimental is done for the validation analysis.(4)In terms of proprioception feedback,combined with the mechanism balancing mechanism and proprioceptive,design the experiments on proprioceptive feedback.Finally,we have done the experiments and analysis aboat different paradigms of motor imagery recognition,to explore the influence on the classification and recognition of motor imagery in the section of no-feedback,positive-feedback,negative-feedback,and forged-feedback.Experimental results show that only the proper feedback can improve the classification rate effectively.However,improper feedback may lead to reducing classification,and even lead to BCI not carrying out.