Abstract: In the case of uncertain returns, in order to minimize the cost of the reverse logistics network in e-commerce enterprises, a multi-level multi-site mixed integer programming model is established. The target area is gridded. The distance between adjacent alternative sites and different levels of alternative sites is calculated by the distance formula on the ball. The Monte Carlo method simulates the return quantity, and designs a genetic algorithm with double chromosome coding to solve. Numerical examples verify the feasibility of the model and finds the location of the delivery sites and the return processing center when the total cost of the reverse logistics network is minimum. There is a corresponding relationship between the location of the delivery sites and the amount of returned quantities. Studying the amount of returned quantities from customers will optimize the location of the logistics network.

Key words: reverse logistics, return recycling, Monte Carlo, MILP, genetic algorithm

摘要： 在退货量不确定情况下,为追求电商企业逆向物流网络成本最小化,建立了多层次多站点的混合整数规划模型。将目标区域进行网格化处理,通过球上距离公式计算相邻备选站点、不同层次备选点间的运输距离,使用蒙特卡罗法模拟退货量,并设计了双染色体编码的遗传算法进行求解。算例验证了模型算法的可行性,并且得出了逆向物流网络成本最小时,快递站点和退货处理中心的选址,快递站点的选址与退货量大小存在对应关系,研究客户退货量会使物流网络选址得到优化。

关键词: 逆向物流, 退货回收, 蒙特卡罗, 混合整数规划, 遗传算法