Simulating Driver's Behavior And Modelling For Critical Driving-Reliability In The Traffic Environment

This thesis refers to the topic"Research on Driver's Traffic Safety Reliability Based on The Traffic Physiological and Psychological Characteristic"in the Doctorate Fundation Project of China Education Ministry, used the ADSL driving simulator to do test and simulation on driver's reaction behaviors in critical scene under virtual traffic environment, and compartmentalize the pressure states of driver-vehicle system based on turning and brake critical reaction time. At the same time, this success percentage of avoiding traffic accidents is used to analyse transfer probability and reliability under different states. This realized the modeling of driver's reliability and the forecasting of reliability, and made a good applying base to distinguish single driver's adapting ability(especially critical driver's reliability).The following is the main conclusion of research or the main work:(1)When taking the consideration of the whole safety project of human, vehicle and road, driver's psychological and behavior characteristic must be concerned. Driver as leader in traffic system plays the most important role in traffic system. The velocity of running vehicle is continuously high and bald on the high way, which keeps driver in continuously high driving condition. Driver's physiological and psychological characteristic, such as apperceive slow, misjudgment, fatigue and blind follow etc are all factors ,which have influence of driving safety on high way.(2)Used the developed SHENCK chass testing system to test the velocity and slipping capability of simulative vehicle in ADSL vehicle driving simulator; Used ADSL to test and simulate driver's critical reaction behavior under virtual traffic environment. This provides a method to research on driving behavior.(3)With reference to questionnaire and analysis& testing on depth perception, attention centrality and attention assignment, designed the scene of nearly accident avoided by emergent turning, analyzed driver's critical turning reaction ability and established multivariate linear regression model.(4)Designed the scene of nearly traffic accident avoided by emergent brake, established nerve-net model when driver brakes in the process of avoiding tail-crashing ,emulated 4 output processes related with brake and acceleration and get the reaction time of driver's critical brake. The model can distinguish driver's brake behavior well.(5)Established 6-state driver–vehicle pressure system reliability models; with reference to the reaction time to turning and brake, had research on the transfer probability and reliability under different pressure states.(6)Use driver's behaviors reliability Markov Model and driver-vehicle system reliability model under pressure together to establish driver-vehicle system critical reaction reliability model and use Monte Carlo method to do emulational forecasting analysis on critical reaction reliability of driver-vehicle system.The main innovations in this paper:(1)Put forward the arithmetic to analyse brake behavior using nerve-net, and established the model of driver's brake behavior. Input the relatively position, velocity, acceleration, and angle to the nerve-net for training, then driver's critical brake state(4 output processes related with brake and acceleration) is gained and can be used to distinguish the driver's critical brake behavior. It indicates that the arithmetic can distinguish driving behavior very well.(2)Used Markov theory to establish 6-state driver–vehicle pressure system reliability model. Put forward the way to compartmentalize the pressure states of driver-vehicle system based on turning and brake critical reaction time, and success percentage of avoiding traffic accidents. Had a research on driver's reliability and transportation probability of relative status with reference to the reaction time to turning and brake, and success percentage of avoiding traffic accidents. It indicates that this model can be used to analyse the driver–vehicle pressure system reliability model very well based on turning and brake critical reaction time.(3)Put forward the arithmetic of using Monte Carlo method to forecaste critical reaction reliability of driver-vehicle system, and the arithmetic is good at the research on relation between safe operating time and reliability. It indicates that the arithmetic made a good research base for distinguishing single driver's adapting ability(especially driver's critical reliability).