Study On Signal Processing And Application Based On Hybrid Brain Computer Interface

In the global environment of aging population,increasing number of disabled people and brain disease patients,it is an urgent problem to help these special groups who can not take care of themselves to improve their quality of life.Brain computer interface(BCI)technology bypasses the normal brain information transmission channels such as human nervous system and muscle,and directly establishes the connection pathway between brain and external equipment.This technology provides a new manipulator control method for special people and special environment scenarios.Based on the Hybrid brain computer interface(HBCI),this dissertation analyzes the frequency domain characteristics of EEG and the time-domain characteristics of EEG,extracts the corresponding feature parameters and classifies them by different algorithms,and finally realizes control method of manipulator based on HBCI by EEG and EOG.In this paper,the H10 C headset is selected as the signal acquisition device,and EEG and EOG experiments are designed.The experimental of EEG and EOG have the problems of baseline drift,noise interference and EOG artifacts.The value filtering is used for baseline correction.The wavelet hard threshold filtering is used to remove the power frequency and high frequency interference,and second-order blind identification algorithm based on sample entropy is used to remove the artifacts.In order to extract the characteristic parameters that can characterize different blink movements,this paper extracts the features of EEG and EOG signals from time domain,frequency domain and time-frequency domain.The amplitude,mean value,standard deviation and correlation coefficient of EEG and EOG signals were extracted as eigenvalues in the time domain,and the waveform differences between the electrodes caused by different behavioral tasks were compared.In the frequency domain,the AR model is used to extract the power and unit average power of the EEG rhythm signals,to characterize the changes of EEG rhythm under different behavioral tasks,and the difference of EEG and EOG signals between the user's active blink and passive blink;By comparing the signal-to-noise ratio and mean square error of the EEG signals decomposed and reconstructed by the db4 and sym5 wavelet basis functions in the timefrequency domain,sym5 is chosen as the basis function of the wavelet transform.The wavelet approximation coefficient of the wavelet transform is extracted as the characteristic value of the EOG.In order to compare the EEG signal changes of different behavioral tasks,the wavelet energy fraction of each sub-band of wavelet decomposition is extracted to characterize the EEG parameters.The feature dimension reduction and classification were studied for the extracted EEG and EEG feature values.In order to reduce the number of eigenvalues,the genetic algorithm and BP neural network were used to reduce the dimension of the feature.In the case of ensuring the accuracy of feature classification,the number of feature has decreased from 39 to 21.The random forest algorithm was used to classify the ocular electrical characteristics,and the classification accuracy rate of 90.55% was obtained.The BP neural network classification algorithm and the support vector machine algorithm were used to classify the EEG feature values.The results show that the SVM algorithm can obtain higher classification accuracy(accuracy rate reaches 82.78%).The classification results of random forest and SVM classifier are analyzed and the classification results are corrected.Finally,the overall classification accuracy rate is over 94%.In order to complete different tasks safely,flexibly and efficiently,this paper selects the UR5 manipulator with active safety mode as the actuator of the HBCI.The kinematics model of the UR5 manipulator is established.The forward and backward kinematics of the manipulator are solved.The trajectory planning problem of the manipulator in the working space coordinate system and the shutdown coordinate system is completed.A simulation model of the UR5 manipulator was established to test the reliability and dynamic performance of the manipulator for linear and circular trajectory planning.The robotic arm control interactive command based on the HBCI is designed,and the TCP/IP communication connection between the computer and the manipulator is established,and the HBCI based on EEG and EOG is used to accurately control the movement of the UR5 manipulator.