列车车站停车的启发式自学习算法及仿真  

Heuristic Self-learning Algorithms and simulation of train station stop

作　　者：陈荣[1] 陈德旺[1] 

机构地区：[1]北京交通大学轨道交通控制与安全国家重点实验室,北京100044

出　　处：《铁路计算机应用》2013年第6期9-13,共5页Railway Computer Application

基　　金：北京市科技新星计划(2010B015);轨道交通控制与安全国家重点实验室自主课题(RCS201127001);国家高技术研究发展计划(2011AA0502)

摘　　要：利用北京亦庄线采集的大量实际停车数据搭建了列车仿真模型,基于站内安装的固定应答器提供的定位信息提出了启发式学习算法动态调整控制器输出。通过仿真表明,启发式学习算法相对于传统的PID算法和无学习算法有较强的适应能力。在仿真条件变化的情况下,能够将停车误差控制在±30 cm范围内。对比PID控制,启发式学习算法减少了控制输出在停车阶段频繁变化的次数,延长了制动系统的使用寿命。该方法计算量较小,在实际运用中有着广阔的前景。It was constructed a train simulation model using mass of data collected from Beijing Yizhuang Line, and proposed Heuristic Learning Algorithms to adjust dynamically controller output based on precise position information of fixed balises installed on station. The simulation results showed that the adaptability of Heuristic Learning Algorithms were stronger than PID Algorithm and Non-learning Algorithm. The Heuristic Learning Algorithms could control the stop error in the range of ~30 centimeters with different changed simulation situation. Contrasted to PID algorithm, the Heuristic Learning Algorithms reduced controller change times during the train stop phase. With the performance, the service life of Brake System could be improved. The computation complexity of this Algorithms was lower, and might be applied to practical work in the future.

关 键 词：应答器 启发式学习算法 精确停车 

分 类 号：U231.4[交通运输工程—道路与铁道工程]