Asynchronous Brain-computer Interface For Fusion Of Event-related Potentials And Visual Evoked Potentials

Brain-computer interface(BCI)is to establish a signal pathway independent of the body's muscles and peripheral nerve tissues through the computer,and to directly communicate information between the human brain and external devices.BCI has been widely used in military,medical and life fields.Event-related potentials is commonly used in BCI technology due to their characteristics of easy induction and no training.Asynchronous BCI is a system that can detect the brain state of the subjects and output instructions only in the working state.It has the advantages of free control and not limited by the system time.At present,most of the BCI based on event-related potentials are synchronous BCI or asynchronous BCI with multi-paradigm fusion,resulting in the inability of the subjects to independently determine the time of control instruction output and the need to learn multiple paradigms to increase the control difficulty.Therefore,it is an urgent problem to design an asynchronous BCI system based on event-related potentials to reduce the difficulty of operation.First of all,in this paper,it is found that under oddball paradigm,both event-related potentials and visual evoked potentials could be induced.When subjects are staring at the noncommand output interface,visual evoked potentials could not be detected in the brain signals.Therefore,visual evoked potentials could be used to identify the idle state of the brain.In this paper,an asynchronous BCI system is constructed by simultaneously inducing event-related potentials and visual evoked potentials under a single paradigm in software environments such as C++ and MATLAB and Neuroscan eeg acquisition platform.The specific work is as follows:The second,in order to extract the induced visual evoked potential in the visual evoked interface,two different extraction methods of visual evoked potential are proposed and compared,namely,the superposition then post-frequency domain analysis and post-frequency domain analysis then superposition.The evoked potential extracted by the method of superposition average post-frequency domain analysis has better interpretability and the accuracy of recognizing the idle state of the brain can reach 98.12%,which meets the requirements of the system and provides a foundation for the construction of asynchronous system.The last,to stimulate brain idle state and target detection aspects and discusses the potential of different fusion methods,adopted by the visual evoked potential detecting brain idle and then through the event related potential target instructions and P-FLDA algorithm through improving the visual evoked potential and the integration of event related potential,common instruction brain idle state and target detection.The classification accuracy of the system constructed by the former reaches 91.07%,and the information transmission rate reaches 54.93 bits/min.The accuracy rate of the system constructed by the latter is 93.43%,and the information transmission rate reaches 40.93 bits/min.The two potential fusion methods to detect the idle state of the brain can combine the characteristics of the two potential,avoid the error caused by single potential,and make the system more robust.In this paper,the influence of the value of stimulus onset asynchrony(SOA)on the asynchronous system in the visual induced interface is explored.The asynchronous experiment with several sets of different parameter values shows that the evoked potential varies in time domain and frequency domain with different value.When the SOA is 175 ms,the system can get the highest accuracy for identifying the target stimulus.When the SOA is125 ms,the system can achieve the highest accuracy in identifying the idle state of the brain and the target stimulus in the working state.Through the comparison experiment,the optimal value of stimulus starting asynchronous is found to make the performance of asynchronous system reach the highest.