Research On Key Technologies Of Brain-controlled Keyboard Based On Steady-state Visual Evoked Potential

It is a new human-computer interaction method to use EEG signal to determine the user's intention and build brain computer interface system,which can enable users to control external devices with "brain idea".Among them,a kind of evoked EEG signal called Steady-State Visual Evoked Potential(SSVEP)is popular because of its advantages of many output instructions and high information transmission rate,which has become a research hotspot of multi-objective brain computer interface system.At present,there are still many problems in BCI based on SSVEP.For example,the background noise of EEG signals is strong.Also,the classification performance of the existing SSVEP frequency recognition method is poor under the short time window,and the information transmission rate is low.In order to solve these problems,this paper studies the key technologies of BCI system based on Steady-State Visual Evoked Potential,and implements an online brain-controlled keyboard based on SSVEP.In this paper,we study the processing and classification of SSVEP EEG signals based on user training and without user training.The main research contents of this paper are as follows:1.Without user training,we studied a method of signal decomposition and reconstruction based on multivariate empirical mode decomposition.The signal is decomposed into multiple intrinsic mode functions.We use an IMF component selection method based on prior knowledge and a signal reconstruction method based on grid search.After that,a SSVEP frequency identification method has been performed.Recognition using EEG signals processed by MEMD will effectively improve the classification effect.2.In the case of user training,this paper proposes an Extended Canonical Correlation Analysis method integrated with Task-related Component Analysis(TRCA)as the SSVEP recognition model.In this method,the spatial filter learned by TRCA is used as part of the extended CCA ensemble classifier in the next step,and sine and cosine reference signals,user individual training templates,etc.,are used as references.It has been proved that the model performs well in a short time window and can achieve better recognition accuracy and information transmission rate in a shorter time.3.In this paper,we design and implement an online brain-controlled keyboard system based on SSVEP.In this system,the SSVEP signal is induced by the flashing of the keyboard keys.After the classification of the aforementioned SSVEP recognition model,the user's intention is recognized,then the operation that the user wants to perform is determined,and the spelling of the corresponding character is completed.