Research On Asynchronous BCI Systems Based On Mortor Imagery

Brain Computer Interface (BCI) is a technology set to express thoughts or controldevices through reading the single from human brain, which can take place of nerveand muscle systems including limbs to achieve the interaction between people andcomputers or other devices. BCI is a comprehensive technology which is related topattern recognition, control engineering, signal processing and cognitive psychology.With the development of science and technology, BCI will play an important role inauxiliary of disability, medical rehabilitation, industrial control and other fields.After the analysis of related researches and materials of national and internationalin recent years, the paper takes EEG signal as the source of information to carry outthe research on asynchronous BCI systems based on motor imagery, which are morepractical and first need to distinguish whether the person is imaging or not, and thencontinue to analyze. The main contents of the research are as follows:1. To the issue that the EEG feature of motor imagery is not apparent, differentparadigms respectively for off-line training and feedback training are designed and away of imagination combined with silent count is proposed after discussing in details,which makes people easier to adapt and meanwhile makes EEG features morehighlighted. The experimental results prove that the design of the experimentalparadigm is reasonable.2. To solve the difficult problem of idle state detection in the asynchronous BCI, amethod is put forward, namely that in the premise of two-class classification,through approximate entropy in time complexity and common spatial pattern(CSP)in spatial pattern different features are extracted to make different classifiers whichare combined by vote-based classification method to improve the accuracy ofjudging idle state, and the result of data processing indicates the effectiveness of theproposed method.3. Aiming at the problem of time-delay in an online system of BCI, a method ofreal-time data processing based on moving the file pointer is proposed to ensure real time as much as possible and timely transport the results to the feedback interface orcontrol equipment.Based on the three aspects of research above, a set of asynchronous BCI systemusing brain to control the car is designed, that is, with the UEA-24BZ amplifyingEEG data, the system took the PC as the host computer, using VC++and Matlab toprogram paradigms and process data, while taking STM32as the lower computer toaccept processing results in real time and control the car. Finally, the performance ofthe system in the accuracy and real-time is verified through a series of experiments.