Multi-agent Modeling And Simulation For Traffic Accident Generation

Casualties caused by road traffic accidents remain a serious problem worldwide.The driving behavior of drivers is the root cause of traffic accidents,and the study on the relationship between drivers and accidents is a key research topic for the prevention and control of traffic accidents.In many traffic models considering human factors,attention resource is less described factor.Mental wandering is a kind of process of endogenous distraction in which drivers separate individual conscious activities from driving tasks and transfer them to tasks unrelated to driving.It may change the driver’s ability to perceive the traffic environment,to judge and make decisions and to control the car,thus causing traffic accidents.Therefore,the establishment of a traffic flow model generating accidents by human factors allows the analysis of the relationship between people and traffic flow characteristics from a micro to a macro perspective.Firstly,based on traffic psychology and driver behavior safety,this paper proposes a multiagent system modeling framework,which considers human psychological and physiological characteristics as well as system factors during driving.Then,based on Net Logo software,the human-vehicle integrated vehicle agent unit was constructed by multi-agent modeling method,and the fuzzy logic rule algorithm was used to write the program to realize the behavior output.This paper proposes a traffic flow model of three-lane long straight road Then the simulation parameters are calibrated in the form of questionnaire and data reference to realize the visualization of accident model and traffic evolution process.We simulate the heterogeneous following and lane changing behaviors of drivers,and reflect the driving conformity psychology,that is,drivers will consider the running conditions of surrounding vehicles during the driving process.At the same time,we simulate the process of mind wandering to build the generating mechanism of collision accident and reflect the change of strategy of speed control and lane changing after the accident,and show the traffic congestion and dissipation process,which shows drivers have subjective initiative in perceiving,judging and making decisions under different traffic environments.Finally,the simulation results are compared with the classical model and real data,and the vehicle speed,lane distribution,accident rate,mind wandering and other indicators are analyzed.The verification results are in good agreement with the actual situation.The simulation results show that the speed control strategy of aggressive drivers is changed from relatively high average speed to lower average speed of the other two types of drivers.Under congestion or collision,conservative drivers tend to queue up,resulting in the fastest decline in patience value of lane changing and the largest fluctuation range.Lane changing is conducive to alleviating the imbalance of local vehicle distribution caused by bottlenecks.The higher the concentration threshold,the more beneficial to reduce the incidence of accidents,but the concentration threshold and the number of distractions showed a nonlinear relationship.Collisions make drivers tend to slow down to improve safety,resulting in an average speed of85% of the speed before the collision in a short period of time.The average speed of traffic flow timing sequence does not depend on the concentration threshold,but is related to the collision rate and the degree and persistence of the resulting congestion.Reasonable use of the gap generated by bottleneck sections to change lanes can help speed up the dissipation of congestion.The research proposed the distracted driving caused by the mind wandering,so as to simulate the traffic accident.The results of the simulation described the dynamic process of car following,lane changing,collision and interaction as well as the psychological and physiological state of the human-vehicle agent,and truly and intuitively reflected the traffic flow state of the road network.The model revealed the transmission process and correlation of the influence of driver factors on traffic safety from micro to macro levels,and reflected the significant influence of accidents on traffic flow operation efficiency.It also added a descriptive theoretical framework to the traffic model accident simulation technology,and provided a new basis for better understanding and simulation of traffic flow changes.