In order to solve these shortcomings of optimization model of stage operation plan in marshalling station, such as: model became too large, the applicability was not strong, complex solving methods, algorithms slowly convergence, and so on. The essence of optimal establishment of the stage operation plan was analyzed. The constraint programming (CP, Constraint Programming ) model was set which the accumulation function of the wagon-flow was as model object, the resource of departure trains and disintegration and grouping order was as decision variables, the maximum of the departure trains was up to the overall goal. Solving the model was divided into: using the constraint propagation to get the initial solutions and using the search technology to improve the initial solution constraint broadcast, and constraint programming algorithm engine (CP Optimizer) was used to implement solving the model. The more satisfactory solution was got when the model run on an ordinary PC for about 20 seconds. It showed that the model met the needs of time and space complexity in spot.%为了解决编组站阶段作业计划优化模型规模过于庞大、适用性不强、求解方法复杂、算法收敛慢等缺点,通过剖析编组站阶段作业计划优化编制问题的本质,在此基础上建立了一个以车流累积为模型对象、出发车车流来源及解编顺序为决策变量,出发车最多为总目标的约束规划(CP,Constraint Programming)模型.将模型求解分为:利用约束传播得到初始解和利用搜索技术结合约束转播改进初始解两部分,并用约束规划算法引擎(CP Optimizer)实现模型的求解.实验表明,本模型均能在普通PC机上运行20 s左右就能得到最优的方案,符合现场对算法时间及空间复杂度要求.
展开▼
