Dynamic Risk Analysis For The Road Transportation Of Hazardous Materials Based On Bayesian Network

Road transport of hazardous materials is a typical dynamic process.The risk changes dynamically with the ever-changing passing locations,weather and environment during transportation.In view of these characteristics,it is of great significance to establish a dynamic risk assessment model for road transportation of hazardous materials,analyze transportation risk in real time and quantitatively,and find out the vulnerable links of accidents.This paper takes hazardous materials road transportation system as the research object and conducts dynamic risk assessment based on Bayesian network.Firstly,the system composition and transportation characteristics are described.The system hazards are identified,and the definition and evolution mechanism of the dynamic risk of hazardous materials road transportation are clarified.Then,two Bayesian network modeling methods,hybrid modeling and expert empirical modeling,are adopted to build the risk assessment model respectively.Based on the reality of hazardous materials road transportation,the more suitable model is brought into the instance for further reasoning analysis.The results show that:(i)By combining Bayesian network model with data information such as GPS and real-time observations,the risk of hazardous materials road transportation can be output in real time through Bayesian inference and updating calculation.It can well solve the complex and uncertain situations in the dynamic process,and is of great significance to the dynamic and quantitative risk modeling research for hazardous materials road transportation.(ii)Based on the limited accident data,expert empirical modeling is superior to hybrid modeling on the construction of Bayesian network.(iii)In the case of sufficient data,the hybrid modeling of Bayesian network,which highlights the importance of daily data accumulation in risk assessment of hazardous materials road transport,can provide reference for the construction of risk assessment model based on Bayesian network learning.(iv)The expert empirical modeling of Bayesian network can solve the NP problem of Bayesian network structure learning and the problem of inaccurate parameter learning,and provide a solution for Bayesian network modeling when data is difficult to obtain or data is limited.(v)Risk factors such as fatigued driving,overspeed and populated regions,are the important factors that affect the risk level of system,should be paid more attention in the risk management of dangerous materials road transportation.