Train Driver’s Fatigue Detection Based On ECG And Operational Conditions

Railway is one of the most important transportation now.Train drivers take a long time to work and the working environment is noisy.After long-term driving operations,it is easy to cause fatigue of the train driver that endangers driving safety.This paper studies the relationship between ECG signal feature and operating behavior feature and driver fatigue status,and builds a train driver’s fatigue status detection model on this basis.The main contents of this paper are as follows.(1)According to the characteristics of the driver’s work,the driver fatigue data collection program is designed.In order to not affect the normal operation of train driver when collecting data,an experimental scheme of data acquisition based on driver data acquisition app is designed based on driver’s handheld terminal.The fatigue degree of drivers before and after on duty was evaluated by subjective questionnaire,and the fatigue state of drivers in different experimental stages was determined.The designed data collection experiment provides effective data support for the train driver fatigue detection model built later.(2)The relationship between ECG features and train driver’s fatigue is analyzed.First,pre-process the ECG signal of 30 drivers in different states.Secondly,extract time domain features,frequency features and nonlinear features from them.Then analyze the relationship between ECG features and the fatigue state of train drivers,find the key features that can reflect the fatigue state of the driver.Finally,the optimal time window length for feature calculation is selected.(3)The relationship between operational condition features and train driver’s fatigue is analyzed.First,we preprocess the operational conditions data,and we focus on the constant velocity part of the shunting locomotive.Aiming at the operational condition features of train drivers,a locomotive output power feature method is proposed to extract driver fatigue-related features,and obtain operational condition features with significant statistical differences.(4)According to the characteristics of train driving,a two-layer information fusion structure of feature layer and decision layer is designed.In feature level fusion,ECG features and operational conditions features are the inputs of svm-1 and svm-2,respectively.Dynamic BPA is generated in real time for decision level fusion.The output of the two SVM classifiers is the input of decision level fusion based on D-S evidence to get the train driver fatigue detection results.The two-layer information fusion structure of feature layer and decision layer provides a method for railway driver fatigue detection.In this paper,the special occupation train driver is taken as the research object,and the data collection experiment,qualitative and quantitative analysis are carried out around the ECG features and operational condition features reflecting the train driver’s fatigue,provides theoretical basis,and promotes the evaluation research of railway driver’s fatigue state.Furthermore,the fatigue state detection model of railway driver is built by information fusion method.