Research On High-speed Rail Driver Vigilance Detection Method Using EEG Based On Deep Learning

As the proportion of high-speed railways in people’s travel modes has increased year by year,the safety of high-speed rails has become more and more concerned.Research on safety technology for high-speed rail has become a hot research topic.In this context,this paper combined EEG signals and alertness to detect the alertness of high-speed railway drivers by classifying EEG signals from the perspective of high-speed railway drivers.And put forward a simple and feasible early warning strategy.The main research contents of this paper are summarized as follows:First,the appropriate point was selected according to international standards.A portable eight-channel EEG signal acquisition device was fabricated using Ag-CL dry electrode and Open-BCI development board.And the design experiment collected the driver’s EEG signals in different states.Secondly,the pre-processing operation of the collected EEG signals improves the signal quality.The linear filter is used for preliminary filtering,and then the blind source separation is performed by FastICA algorithm to reduce the correlation between signals of different channels.Then,the separated signal is denoised by wavelet threshold to obtain a reliable highspeed driver EEG signal.The processed EEG signals are organized into training sets and test sets required for neural network training.Then the convolutional neural network and the long-term and short-term memory(LSTM)network were built to train the training set,and the network performance was verified by various evaluation indicators.This paper also determines the correlation between EEG signals at different locations and driver alertness through the attention mechanism.Finally,this paper compares the performance of traditional classification algorithm including SVM and PRCA with the proposed deep learning method in EEG signal classification.And the early warning module is designed.The feasibility of the proposed early warning strategy is verified by experiments.