中文摘要：

传统列车自动驾驶（ATO）控制策略通过提高对目标速度的追踪精度来精确控制工况切换频繁,能耗较大且无法进行全局优化。直接控制列车驾驶的全局ATO控制策略能较好解决传统控制策略的缺陷。列车在自动运行过程中依据不同的全局控制策略,能耗、运行时间误差、停站误差等评价指标均产生变化。由于评价指标存在内部矛盾,不存在所有指标均最优的控制策略。本文提出1种基于动态邻居和广义学习策略的粒子群（ADPSO）优化全局控制策略的算法。该算法通过挖掘线路信息和列车运行信息指导优化过程,以获得在列车安全运行的前提下,满足一定能耗、运行时间误差和停站误差要求的全局ATO控制策略。仿真研究结果表明与其他两种优化算法相比,该算法具有更好的性能。

英文摘要：

Traditional automatic train operation（ATO）control strategy,which realizes precise control of the train by improving the tracking accuracy of the speed of the target,results in frequent shift of working conditions,great energy consumption and impossibility to achieve global optimization.The global ATO control strategy,which addresses the direct control of train driving,is able to remedy the defects of traditional control strategy.Based on different global control strategies,assessment criteria such as energy consumption,run duration error,station dwelling error vary in the process of automatic running of train.Due to the internal contradictions among evaluating indicators,no global optimal control strategy is available for every indicator.The paper introduced a global control strategy of particle swarm optimization based on adaptive dynamic neighborhood topology and generalized learning（ADPSO）.This algorithm guided the optimization process by extracting track line information and train operation information,in order to obtain a global ATO control strategy that meets the requirements for certain energy consumption,run duration error and station dwelling error under the precondition of the safe operation of the train.The simulation results showed that the proposed optimization model,compared with the other two optimization algorithms,provided better performance.