| In recent years,with the accelerated development of domestic urbanization,the use of hazardous chemicals has risen sharply.Road transport is one of the main forms of transportation of hazardous chemicals,and the accident rate of road transport of hazardous chemicals also shows an upward trend with the increase in transportation volume.Due to the high-risk nature of the transportation industry and the diversity of hazardous chemical species,such accidents are often accompanied by bad social impacts,and society is paying more and more attention to such accidents.Therefore,it is necessary to prevent the occurrence of road accidents of hazardous chemicals and reduce the possibility of accidents.It is also a necessary means and measure to reduce the economic loss caused by the accident and maintain the national public safety and stable social development.It is very urgent in China at this stage to study and predict the causes of hazardous chemical road transport accidents and to establish related accident prediction models.Firstly,the related research on the risk of hazardous chemicals road transport accidents and the current status of Bayesian network modeling in the transportation field were reviewed.The different focuses of previous research have been analyzed.The safety management theory and Bayesian network basics to be used were briefly introduced.Secondly,according to a sample of 443 hazardous chemical road transport accidents in China,the variables were analyzed from the time,space,road,environment,and consequences of the accident,and the relationship between the factors and the types and consequences of the accident was analyzed.The Bayesian network is constructed with the framework of people,cars,roads,environment.Accident causation theory and the coupling theory were used as the guiding theory.The 443 accident samples were classified and discretized.Based on Genie 2.0 software,the Bayesian network structure and network parameters machine learning were completed by K2 and EM algorithms.Shen Xiaoyan’s data statistical analysis results were used to verify the scientificity of the model,and then the machine prediction performance indicators were used to measure the model’s prediction accuracy.The Bayesian network reasoning analysis and evaluation of the established model was used to obtain the important factors affecting the risk of accidents in road transport of hazardous chemicals and make recommendations.Finally,the established model was specifically used in the Sichuan-Guangzhou section of the Beijing-Kunming Expressway,and the prediction results obtained by the model were compared with the historical accident data of this section to verify the theoretical and practical value of the model.The transportation risk of hazardous chemicals on the Guangyuan section of the Beijing-Kunming Expressway is predicted by inputting the basic data of the section,the risk prediction results show that the most likely accident for a tanker truck with normal driving conditions under rain or snow or cloudy weather is a rear-end accident with a probability of 39%.It is the easiest when the driver is fatigued and driving in a curve The probability of a bicycle collision accident is 51%,and the probability of a rear-end accident in a tunnel is 38%.The research in this article aims to provide a new method and ideas for the risk analysis and prediction of hazardous chemicals transportation based on the existing statistical analysis methods. |
