Research On Intelligent Driving Algorithms Of High-speed EMU Based On Online Programming

ABSTRACT:With the development of Chinese high-speed railway, the "go global" strategy of high-speed railway is rapidly boosting the "Belt and Road" policy. Compared with manual driving method, the intelligent driving technology has great advantages in operational efficiency and system safety. So it is a new trend to develop automatic driving technology of high-speed railway. Automatic Train Operation (ATO) has been successfully applied in some subway lines, which includes NO.10 subway line in Shanghai and in-build Yanfang line in Beijing. It is a complex control process to meet multi-objective control requirements and to maintain the high speed and running density.Many researchers are devoted to optimizing the off-line speed-distance curve, considering safety, punctuality, energy-efficiency, the riding comfort and so on. Most optimized driving algorithms are mainly based on the off-line speed-distance curve, which is driven by long term operational experiment data. Different from the existed driving approaches, we establish the expert knowledge system based on the control rules and driving data of excellent drivers. The multi-objective control of high-speed EMU can be achieved by using both expert system and online learning algorithms. The research contents are described as follows.Firstly, we establish expert knowledge system by analyzing the driving rules of excellent drivers and data from traction & braking curves. In addition, we use multi point-mass model with single-coordinate as the dynamic model of high-speed EMU.Secondly, based on the speed limit value of monitoring equipments’ output, control reasoning rules are proposed. Posted limit operation is used to reduce the tracking time interval and to take advantage of high speed. Then, dynamic allocated strategies of operation time are designed to satisfy punctuality requirement. In order to realize the real-time control of EMU, we proposed two online updating algorithms, such as exact online programming driving (EOPD) and inexact online programming driving (IOPD). By using the proposed online updating algorithms, the output of intelligent controller is optimized to reduce operation switching times and to improve the riding comfort of passengers.Finally, we use simulink toolbox as simulation platform of proposed intelligent driving method and design the GUI interface to simplify users’ experience. Using the field data collected from Beijing-Tianjin South, we proposed two intelligent driving algorithms, such as EOPD algorithm and IOPD algorithm. The performance of proposed intelligent algorithms can meet the requirement of multi-objective control in high-speed control process. In addition, to verify the adaptability and robustness of the proposed algorithms, we also simulate the intelligent driving algorithms with different parameters’value of high-speed EMU under the complex speed limit.