Study On The Algorithm Of Artifact Imagery EEG Signals And Its Application In On - Line

The Brain-computer interface (BCI) technology that belongs to a kind of Brain-computer/machine interaction control (BCIC/BMIC) has become one of the advancing hotspot in recent years, and Brain-computer/machine fusion control would be the further step for BCI technology. EEG(electroencephalogram) is the typical control signal source of BCI. But some negative features of EEG signal bring great challenges, including low signal-to-noise ratio, low spatial resolution and highly susceptible to contamination of artifact.Brain computer interface based on motor imagery is a kind of very important paradigm in many brain computer interaction control paradigm. For this purpose, this article took the motor-imagery-based brain signal as research object, set up design research for processing algorithm of adaptive brain signals in BCI system, explored more efficient processing algorithm. It was expected that high-quality EEG related to mental tasks could be obtained, in the ways of improving signal to noise ratio, enhancing spatial resolution and eliminating artifact. Comparison test was designed to compare adaptive processing algorithm in this article with others. Corresponding classification accuracy was calculated in the case that the algorithms of feature extraction and pattern recognition remained the same. Finally, the preconditioning algorithm of automatic adaptive brain signal was validated on the real-time online BCI system. Main research contents as below:(1) Research processing method of brain signal artifact, including the main eliminating method of environmental interference and biological artifacts which seriously impacted performance improvement for BCI system. Some important problems were analyzed. Existing research situations, limitations and some important problems to be faced were analyzed, and online self-adaptation, the development tendency for integration the processing methods of BCI brain signal artifacts in controlling was indicated as followed. It was expected to enhance the overall performance of BCI system integrated the variety of adaptive algorithms.(2) Relatively efficient processing method of motor-imagery brain signal artifacts was searched for, and improved algorithm(Wavelet Package Transform+Adaptive Independent Component Analysis)was used into offline analysis of brain signal data in the year of 2008. First using wavelet packet 4 layer decomposition of EEG signals, the extracted 8-30hz frequency band energy of wavelet coefficients, and then after decomposition of EEG signals to make use of the ICA adaptive filter, where the calculated classification accuracy validated effectiveness of subsequent feature detection and pattern recognition of motor-imagery brain signal with this preconditioning algorithm. At last, compared the experimental results with research results at home and abroad, and validated that the processing algorithm in this article had advantages.(3) The adaptive processing method of motor-imagery brain signal artifacts was researched for using in real-time online BCI system. It was expected to enhance the overall performance of BCI system integrated varieties of adaptive algorithms that machine learned collaboratively. But this method had the time delay problem more or less if used for real-time online system on account of heavy computation, and long calculation time. Therefore, solving the time delay problem of adaptive algorithm existing in real-time online system or how to control the delay time in the range of second level became the research emphasis of this article, which could offer technical support to BCI equipment into actual use.