Research On Driving Fatigue Recognition Based On Driving Behavior

The analysis of numbers of traffic accidents shows that the fatal accidents were easily to be caused by fatigue driving.Plenty of Researchs show that the driving performance can be used to monitor the drivers' physiology condition.Questionnaires were taken to survey the direct cause to driving fatigue,the road alignment which was more easily to cause fatigue,the driving behavior characteristics when drivers were at fatigue conditionare.The results showed that:drivers can get fatigue easily when they were lack of sleep,when all the other conditions are consistent,driving on the expressway was more likely to get fatigue than other types of highways,when drivers were in fatigue condition,the steering control ability will decline compared with in normal physiology condition.In the end,the analysis of internal reliability and resurvey reliability of the questionnaires shows the survey result was precise and the conclusion can be accepted.According to the survey results,driving operation behavior(gear,throttle,clutch, steering,and brake) and vehicle characteristics(velocity,acceleration,lane position) was considered as analysis parameter and each of the paramerter was explained.An orthogonal experimental has been designed and the driving duration,radius of horizontal curve and the vertical gradient were considered as factors that may affect the driving performance of the subject,3 levels was selected for each factor. Experiment scheme and the steps were determined finally.Due to the danger of fatigue driving,and based on the reliability analysis of driving simulator,the experiment of this study was taken on the driving simulator platform.The rapid scene modeling method based on LandXML was proposed to build the simulation scene to meet the requirement of orthogonal experiment design. The alignment of expressway was designed according to Design Specification for Highway Alignment of China(JTG D20-2006) and the virtual scene was build.The analysis of various factors after experimen showed that after six hours' driving,the vehicle speed,gear and throttle became centralization,while the brake and clutch didn't change significantly,the change interval of front-wheel angle become small,the lateral deviation value increase gradually.Factor analysis method, was used to the analyze each factor,result shows that the vehicle speed,gear and throttle have the same change characteristic,the related coefficient was high between them,accordingly,build a new factor "integrated operation factor",then using integrated operational factor,front-wheel angle and transverse offset as fuzzy comprehensive evaluation analyze factors,divide the driver fatigue for 3 fatigue degree(normal,critical,fatigue).By analyze the three factors above,determine the degree mark and the weighing,the fuzzy comprehensive evaluation model of fatigue degree based on driver behavior was established.To verify the model,results show that,16 of total 18 drivers when in normal and critical state was identified successfully,and identification of fatigue missed 3 drivers of total 18.Therefore,the fuzzy comprehensive evaluation model is established with certain credibility,and can be used for fatigue driving monitoring and warning.