The Control Method Research For Electirc Wheelchair Based On EEG And EOG

Electroencephalogram (EEG) is a rhythmic biological signal caused by barinneuron activity which contains the relevant information of the human thinking activity.Eyes are the windows of the soul. Electro-oculogram (EOG) is formed by a potentialdifference of the eye between the retina and cornea, which is caused by eyemovements and closely related to the brain thinking activity. How to effectivelyextract the information of EEG and EOG signal, and then develop a rehabilitationauxiliary device controled by EEG and EOG to serve the limb disorder, has become ahotspot in biomedical, BCI, rehabilitation engineering and other fields.Supported by the National Natural Science Fund (61172134) and according tothe requirements of the topic”The Control Method Research for Electric WheelchairBased on EEG and EOG”, this paper describes the research situation of human-computer interaction at home and abroad, introduces the physiological basis of EEGand EOG, analyses the pre-processing, feature extraction and pattern classificationalgorithm of EEG and EOG signal, and finally discusses the control method forelectric wheelchair based on EEG and EOG. The innovation achievements of thispaper are as follows:(1) In this paper, an EEG de-noising method based on EEMD and improvedadaptive threshold lifting wavelet is proposed aiming to eliminate the noise mixed inthe EEG signal. First, noisy EEG signal is decomposed into several IMF componentsby EEMD algorithm. Then some IMF components dominated by noise are extractedthrough the autocorrelation function characteristic method. Next, the noise iseliminated from the extracted IMF components by improved lifting wavelet. Finally,the reserved low-frequency IMF components are reconstructed with high-frequencyIMF components after denoising to obtain the de-noised EEG signal. Results showthat, compared to the traditional lifting wavelet soft threshold algorithm and improvedlifting wavelet soft threshold algorithm, SNR and RMSE of the new method haveimproved significaltly, which laid a good foundation for feature extraction and patternrecognition of EEG signal.(2) Considering that Common Spatial Patterns (CSP) is easy to miss some keyfeatures of EEG frequencies, an EEG feature extraction method based on wavelet packet transform and CSP is proposed, which take the frequency characteristics of theEEG, the relationship of lead and the good frequency characteristics analysis ability ofwavelet packet transform into account. First, the original EEG signals of C3, C4andCz lead are decomposed by wavelet packet. Then four frequency bands similarto,, and wave are selected. Next, the selected bands are reconstructed bywavelet packet to obtain four rhythm components of EEG signal. Finally, the fourrhythm components of the C3, C4and Cz lead are regard as the input of CSP filter forfeature extraction. Results show that the classification recognition rate of EEG featureextraction method based on wavelet packet transform and CSP is better than thetraditional CSP algorithm.(3) In the stage of EEG pattern classification, a pattern classification methodcombined Support Vector Machine (SVM) and D-S evidence reasoning is proposed.First, the EEG and EOG signals are classified respectively by SVM. Then the D-Sevidence reasoning algorithm is used for decision level fusion. Finally, theclassification results after fusion are mapped into the command to control themovement of electric wheelchair. Results show that the recognition rate of EEG andEOG decision fusion is obviously higher than the pure EEG signal and slightly higherthan the EOG signals only.(4) An electric wheelchair experimental program based on EEG and EOG signalsis designed. The visual stimulus software is improved, which can simultaneouslycapture the motor imagery EEG signal and saccadic EOG signal of subjects under asingle experiment.The EEG and EOG signal preprocessing, feature extraction andpattern classification is completed, and the classification results are mapped intocommand to control the electric wheelchair forward, turnning left or right or stop.