Abstract: Railway freight safety is the core content for railway freight quality. Currently, the speed of freight train has been greatly improved, the products transported have been enriched and the total amount of products has been increased, thus, the railway freight safety risk management facedmore challenges. BP nerve network algorithm has the advantages of rapid convergence, accurate calculation, and weakening the influence of human factors etc. a proper evaluation index system for risks was set up according to the potential dangers existing in railway freight safety. It quantified the sample data and reduced the data subjectivity by using the fuzzy algorithm and AHP methods, and carrying out the risk assessment and reached the corresponding conclusions via setting up BP neural network model. Taken various data of Fengtai Railway Freight Center as an example, the results show that the accuracy of predicting the result can be guaranteed by adopting the safety risk assessment model based on BP neural network.

Key words: railway freight, safety risk assessment, analytic hierarchy process (AHP), fuzzy algorithm, back propagation (BP) neural network