The Research Of Controlling Method Of Power Wheelchair Based On EEG

Electroencephalogram (EEG) signal reflects the activity of brain cells, which hasimportant research value in the fields of neuroscience, brain cognitive, clinicalmedicine and rehabilitation engineering. EEG contains a lot of physical andpsychological information, and the signals can reflect the consciousness andbehaviour of human. Therefore, the analysis of EEG can express people’s willing ofmovements, so EEG can be used to control auxiliary devices and solve the problemsof the disabled’s difficulties in movement. The research of brain-computer interfacetechnology which is related to this purpose has become one of the valuable directionsin the field of biomedical engineering.The EEG signal’s acquisition, preprocessing, feature extraction and patternrecognition methods are studied in this paper. This paper analyses the method ofcontrolling power wheelchair by motor imagery EEG, and transforms the drivinginterfaces of Patrafour power wheelchair, develops wheelchair’s control experimentalplatform based on EEG, then executes some test experiments of controllingwheelchair by EEG on this platform. The main research work and the innovation ofthis paper are as follows:(1) In the aspect of EEG’s preprocessing, this paper proposes a method ofremoving artifacts from EEG based on CuBICA. First the EEG which mixed withEOG(Electro-Oculogram) signals were denoised by wavelet package analysis, aftercentering and whitening, the EEG signals which still contained EOG was separated byCuBICA algorithm. The cross correlation coefficient of the separated signals showthat CuBICA algorithm can efficiently separate EOG from EEG, and get pure EEGwhich is useful for next process.(2) In the phase of EEG’s feature extraction and pattern recognition, this paperproposes a recognition method based on permutation entropy and support vectormachine to classify the motor imagery EEG. Several kinds of standard experimentsare designed in this paper to acquire tester’s EEG. C3、Cz and C4 channels’signals areused to control wheelchair. EEG should be preprocessed, and the permutationsentropy is calculated to constitute a three-dimensional vector feature respectively, thenthe support vector machines is used to classify the type of motor imagery EEG. The test experiments prove that permutations entropy is more suitable than sample entropywhen they are used as the feature parameters for motor imagery EEG.(3) According to the requirement of transforming control system of powerwheelchair, this paper designes a circumscribed ATmega16 singlechip system. Thissystem establishes a connection between PC and ATmega16 based on RS-232 serialcommunication standards, and designes user-level communication protocol inaccordance with the requirements of wheelchair’s movements. It implements thefunction of delivering control commands on wheelchair’s control platform.(4) This paper successfully establishes power wheelchair’s control experimentalplatform based on EEG. The design proposal of this platform includes EEGacquisition part、EEG processing part and motion control part. This paper designesand implementes an expansible brain-computer interface software, and transforms thecontrol interfaces of Patrafour power wheelchair, implements the function ofcontrolling wheelchair’s movements by the communication between host computerand singlechip. At last, the performance of wheelchair’s control experimental platformhas been tested for several times, the test results prove that this system is stable.