Research On High-speed Rail Risk Assessment Model Based On BP Neural Network

With the rapid development of the high-speed rail, people have more convenient means of transportation, additionally giving new vitality to the country’s economic development. As the country’s economic artery, Rail transit’s safe operation is directly related to the safety of the people’s lives and property. Because of various incomparable advantages in performance which road transport, ocean transport and air transport don’t have, such as safety, reliability, efficiency, long distance transportation, low transportation costs and strong transport capacity, environmental protection, transit in most inclement weather, and so on, railway turns to be the main transportation on land.However, due to short time for building, new techniques and facilities adopted and the wide coverage, both the construction and the operation of high-speed rail confronted with lots of risks. Therefore, in order to protect the security of the operations of the high-speed rail, the railway sector inducts the system’s security theory into the high-speed rail management and vigorously promotes high-speed rail’s risk management system strengthening railway risk control efforts, thus, reducing the possibilities of accidents and guaranteeing the safety of the high-speed rail.Mainly on the hazard existing in the high-speed railway, this thesis establishes an appropriate evaluation system and builds a model by the method of the BP neural network. Thus, this thesis uses the model to evaluate the hazard of the high-speed rail and draws the corresponding conclusions and reforming suggestions.Firstly, this thesis identifies the main factors influencing the security operation of the high-speed rail through the fault tree method and establishes corresponding evaluation system. Then this thesis uses the fuzzy algorithm to quantify the data of20sample railways, and then uses the weighted summation method to reduce the subjectivity of data. In addition, for simplifying the input of the network, we improve the convergence rate of the network. Therefore, when the input data has large dimension, this thesis uses principal component analysis to reduce the dimension of the score normalized. Thereby this simplifies the BP network structure and improves the training rate.After reducing the dimension of the index system established, the top15of the obtained railway scores are used as the BP neural network’s training data, while the rest5railway scores as the test data of the BP neural network. The result of the test indicates that the predication accuracy of the risk evaluation model amounts to95%, then the effectiveness of the model is of high proof.Finally, this thesis evaluates the risks of high-speed railway in corresponding conditions respectively by BP neural network model and fuzzy evaluation method.Then, the two evaluation methods are analyzed and compared in details. Additionally, we can analyze the main factors affecting the safety of high-speed rail based on principal components derived from principal component analysis. Furthermore, limited manpower and material resources can be put into these main factors’control and treatment, just as the best steel blade should be used in the key points, thereby making the high-speed rail safety management more targeted.