The Controlling Method Research Of Prothesis Based On BCI

Brain-computer interface (BCI) is a hot research area in recent years. EEG is a typical low-frequency, complex signals. People have different EEG signals to different sensory, motor or cognitive activities, through the feature extraction and classification to achieve certain control purposes, such as controlling prosthesis. The artificial limbs based on electroencephalograph are systems for controlling or communicating with other electronic devices by human intentions. The feature extraction and classification play key roles in the research of artificial limbs controlled by EEG.EEG signal processing is the core of the prosthesis control system of the BCI based on hands motor imagery. The accuracy and processing speed are directly related to the system performance and the key to the success of the system. The focus of this study is based on common spatial pattern (CSP) algorithm and wavelet analysis of the preprocessed EEG signals to extract features and the Fisher linear classifier and support vector machine (SVM) to make pattern recognition.First, independent component analysis(ICA) is used to preprocess EEG data removing components like electrophysiological signals and other interference signals, then after removing artifacts from EEG data, apply the wavelet analysis and CSP algorithm to extract features, Finally apply the designed Fisher linear classifier and SVM to classify the features of left and right hand motor imagery into two categories. Classification results show, using CSP and SVM, the accuracy of the 2008 competition data is up to 100%.In this thesis, through the offline analysis and processing of the BCI competition data, the results are passed to MSP430F149 microcontroller by the serial communication means, using the knowledge of circuit design to design the external control circuit, the control of prosthesis can be achieved by writing system tasks.