Research On Decision-making Of Driver Fatigue Based On The D-S Theory

Fatigue driving is the main cause of traffic accidents. Making decision research of the driver’s fatigue state and warning the fatigue in time is of great significance to reduce the traffic accidents caused by fatigue driving. According to the research status and practical application requirements of fatigue monitoring technology,this topic carried out the simulation fatigue experiment,studied the relationship between human physiological signals and driver fatigue by the method of combining experimental research and theoretical method,the specific research contents are as follows.1. Based on simulation driving experiment platform, the topic collected driver’s back pressure, blood oxygen saturation,heart rate and other characteristics of the data in real time synchronization, established the fatigue criterion by the method of combining the subjective assessment of drivers,objective fatigue assessment of others,and PVT test,to classify the collected data under the two states of fatigue and awake.2. A respiratory signal extraction method based on plane back pressure sensor without instrument wearing driving. First selected the maximum pressure point of 256 back pressure test points for the cubic spline interpolation processing,to do correlation analysis with respiratory signal extracted by pulse blood oxygen,we can get the conclusion that back pressure signal contains respiratory signal. This method ensured the accuracy of the respiration signal,and provided an important reference basis for the fatigue monitoring without instrument wearing driving.3. With statistical methods,the topic studied the variation rule of characteristics of the respiratory cycle,respiratory amplitude,heart rate,pulse blood oxygen under different driving condition,adopt the method of single factor experimental design to study the difference between the average and standard deviation of the characteristic parameters in different time window,selected the best time window of the fatigue characteristic parameters.4. Research on the most effective fatigue characteristics of drivers based on rough sets. According to fatigue characteristic parameters under the best time window,a decision table corresponding fatigue properties is established,and based on the discrete breakpoint rule of information entropy,data is discrete and discernibility matrix is established,to do the attribute reduction of fatigue characteristic parameters and obtain the most effective attribute set of fatigue decision. This avoided repeated extraction of attribute values of multiple source signals,and improved the accuracy of fatigue driving determination.5. Decision information fusion based on D-S evidence theory. First through the most effective physiological characteristic data,fuzzy neural network is trained,and related experts’ experience judgment and collected fuzzy information are integrated into the process of fatigue decision-making,and then the output results are normalized to the basic probability function, which was combined by D-S to determine and identify the fatigue state. The method,which first used fuzzy neural network to extract the information characteristics of fatigue driving and then used D-S to fuse,makes full use of the characteristic information,at the same time avoids the one-sidedness and fuzziness of fatigue identification using single fatigue driving characteristics to analyze.