The Optimized Design Of Portable Device For EEG Signals Recording And Its Application In Brain-controlled System

Brain-computer interface(BCI)provides a new artificial output means for the central nervous system(CNS).Brain-robot interaction(BRI)system,which developed from BCI,has potentially broad application in the fields of medical rehabilitation,smart home and entertainment.And the optimized portable device for multi-channel EEG signals recording is conducive to expand the application scenarios of BRI system.Due to performance of the BRI system is easily affected by user's mental state in practical applications,it is of great significance to improve the recognition accuracy of the brain-controlled system under the fatigued state.Aiming at solving problems mentioned above,a portable device for multi-channel EEG signals recording based on MCU of ARM named Brain40 is optimally designed in this dissertation.It can record 40-channel EEG signals with the sampling rate of 500 Hz,and has the comparable performance with commercial NuAmps amplifier.In order to evaluate its performance,both two amplifiers are used to record EEG signals.In the steady-state visual evoked potential(SSVEP)-based experiment,the recognition accuracy of SSVEP signals recorded by Brain40 is 84.48%,which is 0.09% lower than that by NuAmps amplifier.In the experiment of fatigue driving,both devices are capable of identifying fatigue state via multilayer limited penetrable horizontal visibility graphs with an accuracy of more than 90%.The results show that Brain40 has reliable performance in recording the EEG signals during the experiment of fatigue driving and SSVEP-based brain-controlled system.After classifying the EEG signals recorded in the SSVEP experience during normal and fatigued states,the results show that the recognization accuracy of SSVEP is greatly reduced in the fatigued state.In order to improve the SSVEP classification accuracy during fatigued state,a method based on multivariate empirical mode decomposition(MEMD)is used for analyzing SSVEP signals.Furthermore,in order to reveal the fatigued behavior in the SSVEP experiment,multivariate multiscale sample entropy(MMSE)is adopted to analyze the multi-channel EEG signals under normal and fatigued states.We find that the MMSE values in fatigued state are lower than that in normal state,which reflects the increase of regularity in SSVEP signals during the fatigued state.That is,a greater synchronization of neural assemblies is required to realize cognitive impairment when fatigue happens.