摘要
基于一定的边际假定、定义及其定理,将铁路结点站间集装箱班列开行方案(BCTFP)箱小时消耗最少的优化目标描述为线性阶跃函数,得到BCTFP的优化模型。在模型中,每支非零箱流均对应1个线性等式约束,且每个约束条件之间没有任何交叉。将该模型改造为不含约束条件的0-1二层线性规划模型:上层规划的目标为箱小时节省最大,下层规划的目标为在给定决策变量条件下的沿途改编箱小时消耗最小。按照适应性遗传算法的思想确定遗传策略,采用协同多群体遗传算法,以有效地克服由于问题本身具有强基因关联和超多峰性质而带来的模式欺骗问题,设计相应的遗传算法。通过对算法每个环节计算复杂度的分析,得到该算法的整体复杂度为O(αn3lnβn2),说明该算法是收敛于全局最优的有效算法。
Based on stated boundary assumption, definition and theorem, the thesis described the minimum spending of container hour as linear step function to get the optimization model of Block Container Trains Formation Plan （BCTFP）. In the model, each non-zero container flow corresponded to a linear equation restricton without any intercrossing between each other. After that, the model was transformed to a 0-1 Bilevel Linear Programming （BLP） one. The objective of BLP upper was the maximum saving of container hour, yet with the presented decision-making variables that of BLP lower was the minimum container hours spent on the way adaptation. The adaptive GA was adopted to establish the hereditary strategy and cooperative multi-colony GA was applied to overcome the scheme deceiving resulting from the strong generelated and many-extremumed character possessed by the problem itself. Then, we designed corresponding GA and presented the process. Finally, through the analysis of calculation complexity of each step of the algorithm, the whole complexity of the algorithm was obtained as O（αn^3 1nβn^2）, which showed that the al- gorithm is an effective algorithm that converges to the global optimization solution.
出处
《中国铁道科学》
EI
CAS
CSCD
北大核心
2008年第1期97-101,共5页
China Railway Science
基金
铁道部科技研究开发计划项目(2002X019-B)
关键词
集装箱班列
列车编组计划
结点站
箱小时
方案优化
优化模型
遗传算法
Railway block container train
Train formation plan
Railway network container freight station
Container-hour
Scheme optimization
Optimized model
Genetic algorithm
