Application Of Data Mining For The Mechanism Of Hazardous Materials Road Transport Accidents

With the continuous development of China’s economy,the process of industrialization of society has been accelerated,and China has become the world’s largest producer of chemicals.With the booming development of production and sales and other fields,the chemical logistics born along with it has also increased rapidly.However,due to policy constraints,geographic conditions,non-standardized technical conditions,non-uniform equipment,information is not interoperable and other factors,railroads,waterways and other modes of transport are not fully utilized,resulting in most of the hazardous materials need to be transported by road.Due to the physical and chemical characteristics of hazardous materials themselves flammable,explosive,toxic,etc.,in the process of transportation once the accident may cause more serious casualties,property damage and environmental damage,social impact is bad.Therefore,it is of great practical significance to study the mechanism of hazardous materials road transportation accidents.Based on the above background,this paper firstly analyzes the data of hazardous materials road transport accidents based on statistical methods to reveal the accident characteristics.Secondly,association rules are used to mine the factors and sets of factors that are strongly associated with the occurrence of hazardous materials road transport accidents of different severity.Then the performance of AdaBoost,XGBoost,Support Vector Machine(SVM),Multilayer Perceptron(MLP),and Stacking models in predicting the severity of hazardous materials road transport accidents is compared based on Accuracy,Recall,Precision,F-score,and Area Under ROC Curve(AUC).Finally,the factors influencing the severity of hazardous materials road transport accidents in each of the seven regions of China are analyzed based on the best prediction models.In total,the following main research results were obtained.1)The 10 frequent itemsets related to hazardous materials road transport accidents were found to be,{road type: highway},{hazardous materials type: flammable liquid},{season:summer},{fatigue status: no-fatigued},{moving status: straight},{segment type: normal roadway},{direct accident form: 2-vehicle rear-end},{moving status: turning},{road alignment: flat and straight} and {direct accident form: rollover} using association rules.2)Among the prediction models,the stacking model performed the best.3)In the process of predicting accident severity in different regions based on the stacking model,it was found that there were certain differences among regions in terms of the features that had a significant impact on accident severity,and the impact of the same feature on accident severity varied among regions.The results of the study provide a theoretical basis for emergency management departments,transportation departments and public security traffic management departments in each region to develop scientific and effective supervision methods,accident prevention measures and rescue programs.At the same time,also to hazardous materials road transport enterprises for drivers,escorts,loaders and managers to develop targeted driving skills and safety awareness training,transport vehicles and equipment selection and maintenance,transport route selection to provide the basis.In general,for the prevention of hazardous materials road transport accidents or reduce casualties,property damage and environmental damage,to improve road transport safety has a certain degree of help.