高速铁路接触网悬挂系统维修计划的优化研究  

Research on Optimization of Catenary System Maintenance Schedule for High-speed Railways

作　　者：李雪[1] 吴俊勇[1] 杨媛[1] 严翔[1] 刘晓民[1] 徐伟燕[1] 

机构地区：[1]北京交通大学电气工程学院,北京100044

出　　处：《铁道学报》2010年第2期24-30,共7页Journal of the China Railway Society

基　　金：国家自然科学基金资助项目(60674005);北京交通大学科技基金项目(2006XM032)

摘　　要：将高速铁路接触网系统的维修方式分为3类,根据不同维修方式下系统的可靠性和维修费用不同的特点,建立不同维修组合下接触网系统的动态可靠性模型和维修费用模型。为实现提高接触网系统的可靠性并降低维修费用的目标,提出一种混沌自适应进化算法(CSEA)来求解这一多目标优化问题。该算法的混沌初始种群算子提高了初代种群的多样性,分组选择策略保证各代有一定数量的劣势个体能参与进化,自适应遗传算子增加了劣势个体的交叉和变异概率,从而避免算法早熟,增强了算法的全局搜索能力。计算结果表明,CSEA在种群多样性保持和帕累托(Pareto)最优解收敛方面均优于流行的NSGA-Ⅱ多目标算法。采用CSEA算法得到的优化维修计划,可显著提高接触网系统的可靠性,也大幅度降低维修费用。将本文多目标优化算法与传统的单目标优化算法进行比较,验证了多目标优化算法的优越性。The dynamic reliability model and maintenance cost model of the catenary system along high-speed railway lines have been established at first in this paper,where all the maintenance operations have been categorized into three types and different types of maintenance operations lead to different reliabilities and costs.The optimization of the maintenance schedule is to achieve the maximum reliability of the catenary system and the minimum maintenance cost at the same time.To solve this typical multi-objective optimization problem,an advanced Choas Self-adaptive Evolutionary Algorithm,called CSEA,has been proposed in this paper.The chaotic initial population helps to improve the initial diversity.The grouping selection strategy has been suggested to give the dominated solutions more chance to enter the mate pool,and a self-adaptive genetic operator is adopted to give the dominated solutions higher possibility to cross and variation.All of the above help to avoid prematurity and enhance the global searching ability of the algorithm.Simulation results show that,CSEA outperforms NSGA-II in terms of diversity-preservation and converging closer to the pareto-optimal set at one-run time.At the end of this paper,the comparison of CSEA with the single-objective optimizing algorithm has been done,showing that CSEA can obtain the limit boundary of the single-objective optimizing algorithm and possesses advantages over it in algorithm stability and effectiveness.

关 键 词：多目标进化算法(MOEA) 混沌初始种群 自适应遗传算法 接触网系统(CS) 

分 类 号：U226.8[交通运输工程—道路与铁道工程] O213.2[理学—概率论与数理统计]