含有连续型随机参数的回收物流网络优化设计

Optimal Design of Returned Logistics Network with Stochastic Parameters of Continuous Distribution

考虑到设计参数的不确定性,建立了单一产品、有处理能力约束的回收物流网络优化设计的二阶段随机规划模型。该模型借助于抽样技术给出了连续型随机参数的有限离散数值,利用混合遗传算法计算并比较了不同网络的建设和运营费用,从而避免了网络设施数量和随机向量维数对模型求解效率的影响。为得到稳健的回收物流网络,利用大样本对计算获得的可行网络进行了评价。考虑到样本随机性的影响,给出了基于随机模型的回收物流网络优化设计步骤。另外,通过算例说明了随机模型的有效性,证实了确定性模型近似处理随机规划问题的不适用性。

Considering uncertainty of design parameters, the paper develops a two-stage stochastic pro- gramming model for optimal design of a single product, capacitated returned logistics network, which gives sets of finite and discrete values of stochastic parameters with the help of sampling technique, computes and compares construction and run expenses of different networks by mixed genetic algorithm, so effectively avoids network facility numbers and random vector dimensions′ negative effect on model calculation efficiency. In order to get a robust returned logistics network, using a larger quantity of samples estimates the feasible networks resulting from time after time necessary optimizations. Taking into account the effect of sample randomness, It presents the design steps to acquire a realistic returned logistics network according to the proposed model. In addition, a useful example illustrates the model＇s validity, also verifies the fact that determinate model has no advantage to stochastic programming problem.