基于鲁棒优化的逆向物流网络设计研究

Research on reverse logistics network design based on robust optimization

逆向物流的流量预测困难,加上各种成本的不确定,其稳定性难以保证。为研究其选址问题,建立了一个以总成本最小为目标的标准模型,并在标准模型基础上加入因无法满足的处理量而产生的惩罚成本。分析出其中可能的不确定参数,进行规范式处理,引用p-robust建立鲁棒优化模型。随机构建一个案例,考虑外界参数不确定,将标准模型和鲁棒优化后的模型分别使用粒子群算法进行求解,得出不同的选址结果,对比表明高度不确定下的鲁棒优化模型的选址决策具有较低的风险。分析各不确定参数对总成本波动的影响程度,得出逆向物流量对总成本波动影响最大,应加以重视。最后考虑网络结构不确定,给出一种在不同不确定等级下的选址方法。

It is difficult to guarantee the stability of reverse logistics because of the difficulty in forecasting the flow of goods and the uncertainty of all kinds of costs.In order to study the location problem,a standard model with the goal of minimum total cost is established,and the penalty cost caused by unsatisfied processing capacity is added to the standard model.The possible uncertain parameters are analyzed and normalized,and the robust optimization model is established by using p-robust.A case is constructed randomly,and the standard model and the robust optimization model are solved by particle swarm optimization algorithm,considering the uncertainty of external parameters,and different locations are obtained.The analysis of the influence of uncertain parameters on total cost fluctuation shows that reverse logistics has the greatest effect on total cost fluctuation,which should be paid attention to.Finally,considering the uncertainty of the network structure,a method of location selection under different uncertainty levels is given.