Research On Fatigue Driving Detection Method Based On Vehicle-mounted Device And Driving Behavior Characteristics

As every Chinese have reached 10000 dollars one year and with the development of science and technology,many people are longing for a better life,so people are passionate on buying cars.The number of cars in China are more lager than before.The sharp increase in the number of vehicles has brought a serious social problem.More and more vehicles accidents are taking out.After researching found that fatigue driving is the top three cause of traffic accidents.Therefore,studying the causes of fatigue driving,screening objective and small error detection parameters,and exploring efficient,accurate,and real-time fatigue driving detection schemes can help reduce road traffic safety risks,thereby protecting people’s lives and property safety.Based on the analysis of the driver’s head node,heart rate,grip force movement and the driving performance and combined with data fusion technology,under normal and fatigue situation,set up the driving performance and physiological movement characteristic parameter.At the same time,characteristic indexes of drivers and vehicles are collected.The main works are as follows:(1)Choose the experiment method,formulate the experiment plan.Physiological(nodding,heart rate,grip strength)characteristics and vehicle behavior of the experimenters are collected from CHD students,by driving experiment cars on CHD simulated driving platform.The data was processed in advance,and a sample library was finally obtained.(2)The statistical method was used to study the fluctuation of driver’s driving performance under normal driving condition and fatigue driving condition.The changes of steering wheel angle,vehicle speed and acceleration are studied in detail.For the physiological condition of the driver,it mainly analyzes the indicators of nodding,heart rate and grip strength under normal and fatigue conditions.Through the division of driving time,the data at specific moments are collected and the difference level of characteristic indicators is quantitatively studied.The driver’s physiological characteristic parameters are selected after the analysis.They are the HNR,HRVM,HRVSTD,FM and t FSTD.The driving behavior indicators include: SAM,SASTD,AM and ASTD.(3)Summarize the applicable scenarios of data fusion technology,and choose the BP neural network algorithm.Next,the steps of establishing a recognition model are described.Next,the sample set is selected and trained and verified by the neural network toolbox of Matlab software.The average recognition accuracy of 80.9% and 79.2% under normal driving and fatigue driving conditions is obtained rate.This recognition accuracy rate can be used in fatigue driving detection,and also provides a new solution for fatigue detection.