Research On Fatigue Detection Method Based On Driving Behavior

Fatigue driving has become one of the main cause of traffic fatalities, many researchers at home and abroad has done a lot of research on heart signals and driving behavior about fatigue state identi fy.In order to reduce traffic accidents brought by fatigue driving, people in home and abroad get driv er’s physiological signals and heart signals through detection devices to study the state of driving and obtain the results. But the detection devices of getting driver’s physiological signals and heart signal s have serious impact on the driver’s driving the process, thus impeding the application of Health hea rt research in the vehicle warning system. Because of the driving behavior of the signal detection with a non-contact characteristics, obstacles of interfere with the drivers’driving process is removed. With urgent practical needs and sensor technology as well as the rapid development of computer technology, driver behavior by analyzing the signal to detect the state of the driver driving has better prospects.This paper studies the state of fatigue driving behavior based on driving behavior, through changes in driving behavior data to identify the driver in normal driving state, fatigue driving state or deep fatigue driving state. First driver fatigue causes in driver fatigue warning systems research status and development trends at home and abroad is introduced, on the basis of driving simulation chamber the experiment is designed, with the use of multiple sensors collecting driving behavior signals. By analyzing driving behavior of the driver’s steering wheel angle and throttle position changes, using the time window to extract the steering wheel no move time duty, using optimization of no move time duty of the steering wheel with analysis of variance, amplitude of throttle position mean values and throttle wavelet energy spectrum entropy are calculated and, finally no move time duty of the steering wheel angle and throttle amplitude mean time duty are selected as the feature vectors, Fisher linear discrimination algorithm and distance discrimination analysis model are constructed to identify the driver status. By comparing the two models, Fisher discrimination model in driving state identification is better.