Study On Driving Behavior Safety Multiple Attribute Evaluation Methods And Its Applications

Along with the improvement of people's living standards, the car ownership gradually increases. According to the analysis of traffic accident cause survey enacted by Transportation Department at home and abroad, the percentage of car accidents caused by driver's accounts for over 70%. The security of driving behaviors is related to the security of individuals and others. As such, it is necessary to make an in-depth analysis on the security of driving behaviors. At present, overseas research contents concerning driving behavior security assessment are relatively disperse. Limited by the analytical techniques and theoretical tools of driving behaviors, the driving behavior security studies are not mature up to now. The ultimate purpose of this research is to fundamentally solve transportation hazards and improve the driving security level of the whole society. Taking the security of driving behavior as the main research contents, this thesis deeply explores the driving behavior security grades in combination with multi-attribute indicators and variant rule analysis methods during the driving process. This thesis integrates the collection methods of different attribute indicators and accordingly forms a comprehensive characteristic index and assessment model for characterization driving behavior security. The research fruits of the thesis possess significant theoretical meaning and application value, thus offering technological plans and theoretical foundations for transportation security study, car insurance premium definition and other research fields. The main research contents are presented as follows.(1) A review of domestic and foreign research status of driving safety is made, which mainly covers four aspects, including driving safety index captured technology, driving safety characteristic index, driving safety theoretical study and driver safety research and application. In contrast to domestic driving safety studies, studies abroad concerning this field is very comprehensive. However, it has problems in terms of complicated and overloaded characteristic index, less systematic and normative analysis of driving safety. Evaluation methods of driving safety are mostly reflected through special driving conditions, such as fatigue driving, emergency conditions and sub-tasks, in which the driving safety grade evaluation research is rather limited. This paper carries further work based on the existing problems in the studies concerning driving safety grade evaluation system at home and abroad.(2) The realization of the collection of the index of driving safety. Collection method of evaluating driving safety representation is designed based on component analysis of driving safety. Combined with the authenticity of natural driving and safety of special working condition on automobile performance simulator, a driving safety index captured technology was integrated by a real environment visual perception from eye tracker and vehicle CAN communication synchronization acquisition. Also a driver decision and psychological safety index simulator acquisition system is made up by a physiological recorder and a performance simulator, which can synchronous collect psychological, physiological, virtual scene and vehicle moving status. Furthermore, in order to make up the gap left by previous research concerning driver behavior analysis under the environmental awareness, a real vehicle project combined with radar and camera can repeat the construct of the driver driving behavior in the current traffic environment.(3) The safety representation system of environment perception based on vision is analyzed in the paper. We analyzed the visual behavior in lane changing, curve driving and straight driving according to the data of eye movements. Then, we proposed the theory to analyze the safety of lane changing, curve driving and straight driving in Visual Perceptive. We promoted the evaluation method for the safety representation system from multiple perspectives. Also, we figured out the driving safety index in the safety research of driver's behavior.(4) Safety index was analyzed based on the reaction of physiological signal and mental capacity. We designed the simulator condition of Emergency collision avoidance. The data of driving, of answering the phone while driving, of fatigue driving and driving with ACC was collected to evaluate the driver's reaction capacity in emergency collision. The driver's reaction index was confirmed by synthesizing the signal of leg muscles and brake pedal. To evaluate the psychological endurance of the driver in an emergency, we utilized the specificity of heart rate and surface electromyography. Besides, we established standard to evaluate if the driver can react to collision successfully or not.(5) Safety index of handling behavior and state of motion were also analyzed. The road information collected by radar and camera was combined with the handling behavior and state of motion. The index from lateral and longitudinal was established. We promoted the rationality index of emergency brake and lateral deviation index based on lane detection. Finally, we presented the safety index to evaluate the braking action, stability of steering wheel angle, turn signal warning, distance and rate of acceleration.(6) Research on the safety multi-attribute evaluation method of driving behavior was made. Pearson correlation coefficient analysis was carried out to examine the safety evaluation indexes of multi-attribute driving behavior, which were obtained by three links including drivers' visual perception, decision reaction and mental endurance, vehicle manipulation and running status. The processing and refinement of index type, such as the consistency of index type and the non-class quantization, was standardized. The safety evaluation index system of driving behavior was established. The weight of the driving safety evaluation index system was further studied. The comprehensive weight assignment analysis method based on the combination of subject and object was established by the application of analytic network theory, entropy method and maximum deviation method. Fuzzy triangle function based on the fuzzy slogans was analyzed and processed. At last, Technique for Order Preference by Similarity to an Ideal Solution(TOPSISI) was used to evaluate the security level of the subjects in the study, and MADDES safety evaluation model of driving behavior was established.