EEG Signal Acquisition And Feature Recognization For BCI System

The brain-computer interface (BCI) technology is a communication system which is constructed between brain and computer or extern devices and independent of normal peripheral nerve systems and normal muscle system. BCI system use computer and electronical device to gather and analyze brain wave signals under specific environment and tasks,the send the information or command of our brain to the outside world, then will achieve the purpose of controlling external devices.In the fields like aerospace, robotics, multimedia, rehabilitation medical engineering, artificial intelligence, brain cognition, neural feed back training, BCI has a brilliant future.Nowadays, BCI technology has aroused the attention of many scientists from wide fields and has become a new hot researching point in the cross-over field which contains neural science,electronic information and computer science, and has double value in the fields of science and practice.The beginning of researching of BCI is late, the technology is more complex and request much knowledge of many relative fields.So not only the technology itself, but also the practical researching is in the phase of exploring.To make BCI technology out of the laboratory and could be used in real world, there are many problems need to be solved, such as signal acquisition, signal processing, the speed of information transmitting, the choice of experimental paradigm, the accuracy of recognition and also the training of the subjects.Against the problems of current BCI technology, this paper presents the study on EEG signal collection, processing and pattern recognization for BCI system which is based on visual evoked potential P300. Concrete experiments and work is shown as follows:(1)The hardware design for EEG acquisition system. A high performance bioelectrinical amplifier is designed. The skin contact impedance synchronization test circuit, analog notch filter and other filters is added in the circuit. Meanwhile,corresponding test procedures for this amplifier is designed in Labview environment. Finally, digital part and USB transmission part of this system was designed.(2) Software design for EEG acquisition and measurement system. In order to get enough universal property and extensibility, the architecture of the software isn't depend on the specific amplifier and can be regard as a common brain computer interface.The EEG data of visual evoked potentials and somatosensory evoked potentials can be collected, stored and displayed by this software.(3)Signal processing for visual evoked EEG.Independent Component Analyze (ICA) and Principal Component Analyze (PCA) are used. With this method, the 50Hz interference and artificial can be restrained effectively, and noise and signal can be separated in real-time.(4)Real-time feature extraction.A method based on AR model and fewer coherent averages for P300 feature extraction is presented.(5)EEG feature recognition.In this section, the standard data of BCI is used to verify the recognition algorithm, in which the fisher linear discriminate methods and the neural network was used to classify features of the P300.The result showed that these methods can satisfy the requirement of BCI system.This EEG acquisition system and recognition algorithms designed in this paper can be applied to various types of brain-computer interface system, which laid the foundation for the realization of online brain-computer interface system.