Design And Research Of Brain-computer Interface System Based On Steady-state Visual Evoked Potential

The research of HCI (Human-Computer-Interface) is getting more and more attention with the development of the computer science. BC1(Brain-computer Interface) system is a kind of special HCI system, which use the EEG collected from human body to control the computer or other external electronic devices through the link between human body and computer system. The target of BCI research is to design a new assistant device based on EEG for disabled individuals, so as to help them with cinesipathy to communicate with the external environment better. Thus, it has very serious social significance and potential economic value.This thesis designs and implements the BCI system based on SSVEP (Steady-state Visual Evoked Potential). The main research work include the follows:1. Participate in the building of the hardware acquisition platform of BCI system. The EEG signal is very weak and it would be mixed with many kinds of background noise. So, the signal must be amplified and filtered through the pretreatment circuit before the software reading it. The pretreatment circuit contains mainly four parts, they are the fore-end protect circuit, radio frequency filter, middle amplification circuit, common mode rejection circuit and back-end amplification circuit.2. Design and implementation of the BCI system software. The system designs and implements the visual stimulator module and extracts the characters through many digital signals processing methods, for example filter and frequency analysis, so as to control the peripheral equipment like robotic arm. The system software contains five modules and they are signal collection module, EEG display module, signal pretreatment module, signal processing module and signal feature extraction module.3. Experiments and analysis based on the designed BCI system. Several postgraduates are selected to be the experimenters to do experiments of the system and data collection. A lot of original data of SSVEP are gained and the SSVEP signal sample library is built preliminary.4. Analyses the problems existing in the current system, then proposed a new algorithm of sliding windows variance estimation in terms of SSEVP spectrum detection threshold selection problem, which is used for online normalized of windows data energy. This algorithm relieves the difficulty of detection threshold choice to some extent and the experiment result proves the validation of the new algorithm.