基于箱流波动的集装箱中心站直达班列开行方案优化

Optimization of Operation Scheme of Through Train from Container Central Station Considering Container Throughput Fluctuation

摘要:

研究一类契合箱流波动的铁路集装箱中心站直达班列开行方案优化问题。以班列始发集结车小时和沿途改编车小时消耗最小为目标，考虑中心站改编能力和箱流组织方案唯一等限制，利用统计时段内集装箱流日均值，构建基于箱流日均数据的中心站直达班列开行方案优化模型。进而直接利用集装箱流波动数据，构建基于箱流动态数据的中心站直达班列开行方案优化模型。然后开发概率约束的确定性转化方法，并设计基于车流序列优选-更新-接续的元启发式求解过程。最后，设计实验场景对箱流日均数据模型和箱流动态数据模型的中心站直达班列开行方案进行比对。结果表明：集装箱流在±10%、±15%、±20%和±25%波动区间，日均数据模型的稳定性分别为77.2%、72.4%、68.4%和51.6%，动态数据模型的稳定性分别为96.0%、92.8%、88.2%和77.2%，动态数据模型适应性更强。

Abstract:

The optimization of operation scheme of through freight train from container central station considering fluctuating container throughput is proposed. The objective function intends to minimize the total time of container accumulation and train reformation. The model considers some constraints including the station classification capacity， the classification tracks number， and the container flow formation scheme uniqueness. The daily average container throughput is counted， and the formation plan of through freight train from container central station with certain container throughput is obtained. Then， the daily dynamic container throughput is directly adopted to establish the operation scheme of through freight train from container central station with fluctuating container throughput. Afterwards， a method of transforming random constraints to deterministic constraints is developed to deal with the daily dynamic container throughput， and a meta-heuristic based on filtering-updating-connecting wagon flow sequence is given to obtain the container train operation scheme. Finally， the experimental scenarios are given to test the through train operation scheme models with daily average container throughput and daily dynamic container throughput. Under the container throughput fluctuation range of ±10%， ±15%， ±20%， and ±25%， the feasibility measure of the model with daily average container throughput is respectively 77.2%， 72.4%， 68.4%， and 51.6%. However， the feasibility measure of the model with daily dynamic container throughput arrives at respectively 96 .0%， 92.8%， 88.2%， and 77.2%. Obviously， the latter has more advantages.