Research On Driving Fatigue Monitoring And Early-warning Technology Based On The Reaction Time

Driving fatigue has a great impact on the thinking,judgement,reaction ability and performance of a driver,which may reduce the control ability to the vehicle.Laws and regulations prohibit long-time and continuous drive to prevent driving fatigue.But for personal reasons,fatigue also may occur during the allowable driving time.Especially for the drivers of freight transport and passenger transport,the pursuit of economic benefit leads to more frequent accident.Analysis or summary after accident is not enough for safe driving.The transition of being from passive safety management to active pre-warning has its significance.The direct behavior of a driver is slow reaction during fatigue.Slow reaction leads to long operation time.The speed and the lateral position of the vehicle are also abnormal.In this paper,the physiological parameters were thought to be the bridge,to analyze the changes of the reaction time and lane departure during fatigue.The technology about driving fatigue monitoring and pre-warning was developed.The specific research is shown as follows.1)The correlation analysis between the reaction time and the physiological parameter of driver.The electroencephalograph(EEG)and electrocardiograph(ECG)signals were processed by using Fast Fourier Transform algorithm,Welch power spectrum,Haar wavelet transform et al.to extract nine impacting factors.They are the integral of power spectral density(PSD)of α,β,δ and EEG wave,the ratio of the integral of PSD of α and β wave,the ratio of the mean of the power spectrum of α and β wave,the ratio of the sum of the mean of the power spectrum of α and θ wave and β wave,heart rate and standard deviation of RR interval.The grey correlation model was established to analyze the correlation degree between nine impacting factors and reaction time.Results show that the ratio of the mean of the power spectrum of α and β wave has great correlation with reaction time.2)The establishment of the discrimination model of driving fatigue based on the reaction time.The mental state of driver was divided into three levels that were sober(Level 1),slight fatigue(Level 2)and serious fatigue(Level 3).The support vector model(SVM)was used to establish the discrimination model of driving fatigue.The ratio of the mean of the power spectrum of α and β wave and the reaction time were selected to be the inputs of the model,and the fatigue levels were selected to be the outputs of the model.Genetic Algorithm(GA)was used to optimize the penalty coefficient and the parameter of kernel function of the model.The model was verified by experiment.Results show that the accuracy is 86%.Additionally,physiological parameter optimization based on grey correlation analysis can contribute to improving classification accuracy.3)The analysis of the reaction time during driving fatigue.The reaction time was analyzed from different aspects using SVM model.The relationship between reaction time,the ratio of the mean of the power spectrum of α and β wave(α/β)and driving fatigue was analyzed in time domain.The reaction time has a positive correlation with α/β when the driver was in slight fatigue and serious fatigue.The probability density functions of reaction time of all state and serious fatigue state were established,respectively.The change of reaction time was analyzed from aspects of gender and age during different mental state.4)Research on the detection time of driver’s reaction time based on the vehicle running status.The method of image capture of lane departure was designed.The images were processed by using digital picture processing technique.The methods of preprocessing,included graying and mean filtering,were used to highlight the characteristics of images.Sobel lateral edge operator was used in the images segmentation,to extract the edges of vehicle and lane lines.Then the distance between vehicle and lane lines was calculated.The changes of lane departure during different fatigue state were analyzed by using SVM model.Results show that lane departure has an obvious change during serious fatigue state.5)The development of driving fatigue monitoring and pre-warning system.The requirements of the hardware and the software were analyzed and the system was designed.The hardware and the software of the system were designed based on the requirements,respectively.Each function of the system was realized in Android system and ARM 9.The interface of the system was designed.The software and the hardware were integrated to be a whole system.The system was tested by actual drive.The system developed in this paper could monitor the mental state of driver in real time.The device of the system is easy to be vehicle-mounted.This paper provides a new method for safety and early-warning technology of freight and passenger transport.