Intelligent High-speed Train Operation Based On Ensemble Classification And Regression Tree Algorithms

ABSTRACT:With the rapid development of the scale of high-speed railway in our country, the operating speed and traffic density dramatically increase in recent years. Manual operation now is the main way of high-speed train operation, which can not meet the basic requirements for safety and efficiency. Based on computer technology, automatic train operation system not only can operate the train by accurate parking and arriving in on time, but also can ensure the comfort of passengers and reduce the energy consumption. Therefore, researching for better high-speed train operation models is very important to ensure the safety and efficiency of the operation of high-speed train. Using a more efficient high-speed train operation way is the development trend of high-speed railway in the future.Generally experienced and outstanding drivers can operate the train to the specified location accurately, on time and safely. Through the analysis of the actual manual operation data, an intelligent high-speed train operation model has been built in this thesis. By using ensemble CART algorithms, the driving strategies of the outstanding drivers are summarized and described by intelligent language in order to be recognized by computer automatically to operate the high-speed train. This paper mainly consists of three parts:(1) In order to solve the existing problems and deficiencies of manual operation, we put forward our research objectives:build an intelligent high-speed train operation model, make simulation experiment to the performance of the proposed algorithms with manual operation.(2) The second part introduces the modeling procedure of the intelligent high-speed train operation model. Firstly, we analyze the mathematical model of high-speed train. Secondly, we introduce the ensemble CART, collecting method and examples of driving data of drivers. Finally, we describe the intelligent model by the flow diagram.(3) We develop the high-speed train operation simulation platform based on MATLAB software. The proposed intelligent model is simulated on this platform with actual driving data of Hankou-Yichang high-speed railway. Simulation results show that the train can get to the specified location on time, steadily and consume less energy. The performance of the proposed intelligent high-speed train operation model is better than that of traditional manual operation. In addition, by changing parameters of train model and setting the complex speed limit and complex gradient, the simulation results show that this intelligent model can adapt to all these changes, which verifies the robustness of the intelligent model.