Research Of Brain Computer Interface Based On Steady-state Visual Evoked Potential

BCI is a system which is not depend on the normal brain communication pathway,muscles and nerves.The initial goal to build a BCI system is to bring those motor disabilities with a way that they can communicate with the external environment.This thesis focused on the relevant knowledge about Steady-state evoked potential(SSVEP),and developed an online SSVEP-based BCI system.Specific tasks are as follows:(1)Power spectral density analysis(PSDA)and canonical correlation analysis(CCA)are two mainstream feature extraction method.The thesis improves the two method respectively,provides a new PSDA method which is based on empirical mode decomposition(EMD)and gaussian model,named M-PSDA(a modified power spectral density analysis).This method decomposed the raw EEG signal into some IMF(intrinsic mode function),and eliminated the low frequent interfering signals.The remain signals will be reconstructed and analyzed by power spectral density with gaussian model.Experiment results show that the modified PSDA method has a higher identification accuracy.It raised the accuracy of identifying by 12%,which is up to 85% in 3s time window.Because this method only need one recording electrodes,it fits the situation that the recording electrodes is rare or the data acquisition device is a wearable device.(2)Visual stimulator is important to the BCI system.This thesis compared and chose many stimulus parameters.A visual stimulator with a high degree of accuracy has been provided by this thesis and used in the online SSVEP-based BCI system.The visual stimulator was developed with DirectX application programming interface,Windows accurate counter and the principle of frame timing.(3)This thesis also realized an online signal processing system which connecting signal collection,signal processing and feature extraction.This system is programmed by C# language on the visual studio 2013 platform.This online and real-time BCI system is based on SSVEP use a wearable EEG device called Emotiv,and has some convenience.The application of this BCI system is a football game which has four control command,include Up,Down,Left,Right.In 3s time window,the average recognition accuracy is up to 73.7% and the information transmission rate is up to25.75bit/min.This thesis introduced the structure and design solution of this system indetail.The experiments verified the stability and practicability of it,which laid the foundation of more practical and more versatile BCI system in the feature.