Risk Evaluation for Highway Transportation of “Toxic Gas” Hazardous Materials - Based on a Back Propogation (BP) Neural Network

摘要

Risk evaluation models are developed to improve decision making regarding the safety of highway transportation of “toxic gas” hazardous materials. Based on ANN (Artificial Neural Network) theory and the Back Propogation (BP) method, a new risk evaluation model has been established utilizing the assessment indices of the inherent hazard of transported gas, the traffic accident rate of road sections and the population density of roadside areas. 12 road sections whose risk levels have been determined by a fuzzy synthetic evaluation model are used to train the network to an evaluation error less than 10~(-5). Simulation results show that the proposed model in this paper improves on previous published models and has wide applicability.