考虑多种安全库存策略的选址-库存问题研究

Six Location-inventory Models with Risk Pooling in Two-echelon Logistics System

考虑多种安全设置策略的物流网络的选址-库存问题,不仅是选址、订货、运输和库存的集成优化,还需要考虑多种不同的安全库存设置和转运策略。因此,本文深入讨论了二级物流网络中的六种安全库存设置策略,构建了六种考虑不同安全库存设置的选址-库存模型。在考虑集中设置安全库存时,集中安全库存需要通过转载运输实现,因此需要将转载运输成本引入选址-库存模型之中,使新的选址-库存模型更加科学合理。另外,针对六种新的选址-库存模型,提出了基于个体成本差异分配的遗传算法,迭代搜索选址、分配、库存设置策略的优化组合。最后,通过数据实验验证了模型的有效性:(1)安全库存与转载运输之间存在此消彼长的背反关系;(2)安全库存设置和转载运输策略对总成本的影响取决于两种费率权重情况。本文的研究可以为二级物流网络的选址、订货和安全库存策略集成优化决策提供参考依据。

Location-inventory problem and risk pooling strategy are both very common in logistics network.And location-inventory with risk pooling has obtained the attention of some researchers.Inspired by their research,in this paper,six risk-pooling strategies are combined with traditional location-inventory model and six new location-inventory models with risk pooling are built in a two-echelon logistics network.First,six different risk-pooling strategies are introduced.Risk pooling strategy is composed by safety stock policy and transshipment policy.In two-echelon logistics network,the safety stock could set at upstream regional distribution center or downstream frontier distribution center considering two segments lead time separately or both.As the lateral transshipments will happen when the safety stock is set intensively,then the transshipment cost should be considered in the model with centralized safety stock setting policy.So,the main differences of these models are that they have different safety inventory cost function and different transshipment cost function.However,these different risk pooling strategies could also affect the location and allocation decision indirectly.As the models with risk pooling strategies are non-convex,and they are hard to solve,a genetic heuristic algorithm is proposed.In the genetic heuristic algorithm,the genetic part is set to search the location and the heuristic part is set to do the allocation procedure.Finally,a numerical experimental is presented to prove the validity of our algorithms and models.The results prove both our models and algorithms have good performance.They also show that risk pooling strategy really has great impact on the location-inventory decision.And this impact mechanism related to two important trade-off pairs:1)safety stock cost and transshipment cost;2)safety stock cost or transshipment cost and total cost.Furthermore,the trade-off between safety stock cost or transshipment cost and total cost are depend on the ratio of unit cost of safety stock and transshipment.And the threshold value of this ratio is decided by all the other inputs of the model.