| China already has the world’s longest operating mileage and the highest operating speed railway network.High-speed rail operation safety has become a research focus in China and even in the global transportation field.Drivers are an indispensable part of vehicle active safety technology.As a starting point,this paper proposes a high-speed rail driver’s EEG fatigue detection method based on convolutional neural network based on current emerging artificial intelligence technology.The signal judges its fatigue state.Firstly,this paper uses brain-computer interface technology and adopts Open BCI Cyton EEG acquisition device to collect EEG signals from 20 subjects who have been trained in driving simulation on the high-speed train digital simulation platform of Southwest Jiaotong University National Key Laboratory.Secondly,by analyzing the types of artifacts of the EEG signal and corresponding denoising methods,band-pass filtering and linear filtering are used to remove power frequency interference from the collected data,and then independent component analysis and wavelet threshold denoising methods are used to further remove the EEG artifacts.In order to improve the signal quality,the training data set and the test data set are sorted according to their types.Then,this paper uses the Pytorch deep learning framework to build a Res Net-based Convolutional Neural Network(Convolutional Neural Network)and Mobile Net-based Invertedresidual structured convolutional neural network,and then inputs the pre-processed EEG data to the network.Train network weights,and verify and compare the fatigue performance classification performance of the constructed networks through classifier evaluation criteria.Experimental results show that both networks can effectively classify human fatigue status through EEG data.Finally,according to the driving environment of the high-speed rail driver,an EEG fatigue warning method based on the high-speed rail driver is proposed,which is used to detect the driver’s alertness during driving and implement corresponding early-warning measures to ensure the driving safety of the driver.An experimental simulation is performed to easily implement the warning The expected function of the system is of reference significance to the early warning methods of train drivers. |
