Vehicle Operation Risk Assessment Based On Bayseian Network

The safety of vehicle operation is affected by various traffic factors including driver,vehicle and road environment,and the interaction between these factors is very complex.However,most of the existing vehicle risk assessment methods only consider the limited factors,and their applications are restricted to simple scenarios.Which is based on vehicle dynamics and kinematics,unable to adapt to the complex traffic environment.In view of this problem,this paper combines driver factors with vehicle factors to study the method of vehicle operation risk assessment.Firstly,according to the real vehicle driving experiment,this paper obtain the relevant data of the driver’s vehicle operation and eye movement behavior.The dynamic factors that affect the vehicle running risk have been analyzed,and the probability of the factors.The corresponding threshold of each factor under different risk levels is determined.Secondly,on the basis of considering the factors affecting the vehicle running risk,the target system of vehicle running risk is determined.Vehicle operation risk index system includes vehicle spacing,acceleration change rate,steering entropy,time to deviate from the front view,scan range,scan speed and blink frequency.The corresponding threshold of different levels of risk is identified.Then the Bayesian network structure is established by the knowledge of experts.The static and dynamic evaluation indexes are constructed,and the model can quantitatively describe the vehicle running risk under the influence of different factors.Finally,The posterior probability reasoning and maximum likelihood reasoning are analyzed by Netica.The results show that the Bayesian network model can effectively mine the complex relationship between the risk and the factors,and express the uncertainty between the related variables.The model cannot only predict the probability of occurrence of risk in a given situation,but also can find The key causes and the most unfavorable combination that lead to the risk.The results can provide theoretical support for the analysis of road traffic accident inducing factors in urban road management department and lay a foundation for improving the safety performance of urban road traffic system.