地铁列车舒适度最优的自动驾驶过程建模及求解  

Automatic Driving Process Modeling and Solving of the Riding Comfort Optimal Problem for Metro Trains

作　　者：冯江华 黎向宇 胡云卿 王斌 FENG Jianghua;LI Xiangyu;HU Yunqing;WANG Bin(CRRC Zhuzhou Institute Co.,Ltd.,Zhuzhou,Hunan 412001,China)

机构地区：[1]中车株洲电力机车研究所有限公司,湖南株洲412001

出　　处：《控制与信息技术》2018年第6期32-37,共6页CONTROL AND INFORMATION TECHNOLOGY

基　　金：国家重点研发计划(2018YFB1201600)

摘　　要：乘坐舒适度最优是地铁列车自动驾驶研究中的典型问题。文章在考虑地铁列车的牵引/制动特性、运行准点和精确停车等要求的基础上,建立了列车在站间行驶的乘坐舒适度最优驾驶模型;并进一步提出控制变量和状态变量的同步离散化策略,将决策变量无限维的最优驾驶问题转化为一个决策变量为有限维的凸二次规划问题;采用带预测校正的原-对偶非可行路径跟踪法求解,最终获得乘坐舒适度最优的驾驶策略。最后,文章采用一个地铁列车在站间行驶的实际案例进行了仿真计算,结果表明本文所提出的方法可以有效求解乘坐舒适度最优驾驶问题。Riding comfort optimal driving problem is a typical one in automatic driving system for metro trains. An optimal riding comfort driving model for a metro train traveling between stations was established, which takes the traction and brake characteristics into account and ensures the requirements of safe operation, on-time scheduling, and accurate parking, etc. In order to solve the continuous problem, a synchronous control variable and state variable discretization strategy was proposed. After that, the optimal driving problem with infinite dimensional control variables is transformed into a convex quadratic programming problem with finite dimensional decision variables. Primaldual infeasible path-following algorithm with prediction and correction method was adopted to solve the quadratic programming problem, which can obtain the riding comfort optimal driving strategy by iterative way to approach the optimal solution from an initial guessed point. To verify the effectiveness of the established model and proposed algorithm, a practical example was calculated and simulation results show that the solving strategy and method proposed in this paper can effectively handle the established riding comfort optimal driving problem.

关 键 词：乘坐舒适度最优 自动驾驶 控制变量离散 状态变量离散 凸二次规划 原-对偶非可行路径跟踪法 预测校正 

分 类 号：U231.6[交通运输工程—道路与铁道工程] TP13[自动化与计算机技术—控制理论与控制工程]