B2C配送中心分散存储优化问题研究

Study on Scattered Storage Optimization Problems of B2C Distribution Centers

由于B2C配送中心客户订单呈现出"单笔订单小、品种多、配送范围广、交货期紧"的特点,以SKU为最小存储单元的传统连续存储方式失去了优势.在这种背景下,本文针对现代化B2C配送中心提出考虑货品间关联关系的分散存储优化问题,目标是最小化所有货品对儿的加权最短距离之和,使得不同的货品总能分散存储在拣货员的附近以提高拣货效率.目标里的权重为货品间的关联关系矩阵,最短距离为每个位置上的具体货品与其他位置上其他货品的最短距离.对该问题建立整数规划模型并利用CPLEX标准优化软件进行求解.基于模型求解时间和求解规模的限制,提出两个具有不同适应值函数的遗传算法对该问题进行快速近似求解.实验结果表明,分散存储方式明显优于传统的连续存储方式;适应值函数取差值的遗传算法在求解精度和求解时间上比适应值函数取倒数的遗传算法的性能更好.

As the customer orders of B2C distribution centers take on new characteristics"small order,large assortment,wide distribution range and tight delivery schedules",the traditional continuous storage method that SKU is the smallest storage unit has lost its advantages.Based on this background,a scattered storage optimization problem considering the association relationship between products is proposed in this paper for modern B2C distribution centers.The objective is to minimize the weighted sum of the shortest distances so that different products can always be found nearby.The weight in the objective is the correlation matrix between products,and the shortest distance is the shortest distance between the specific product in a position and the other products in other positions.An integer programming model is established and CPLEX standard optimization software is used to solve it.Based on the limitation of model solution time and solution scale,two genetic algorithms with different fitness value functions are proposed to solve the problem quickly.The experimental results show that the scattered storage method is obviously better than the traditional continuous storage method.The genetic algorithm with the difference fitness value function has better performance than that with the reciprocal fitness value function in solving accuracy and solving time.