Abstract: Considering the problem of shelves location reassignment of Robotic Mobile Fulfillment System (RMFS)under dynamic demand,the dynamic storage reassignment policy based on matching rate between shelf and location was proposed,and mixed integer programming model which reflected the continuous dynamic variation of warehouse layout was built with the objective of maximizing the correlation between shelves demand frequency and distance to picking station.Single-t Exchange Simulated Annealing(SA-STE)algorithm was designed to solve the dynamic shelves reassignment problems.The results showed that SA-STE algorithm could find the solution close to optimal solution found by CPLEX solver in a relatively short period of time,and its advantages were obvious in large-scale problem.Compared with the traditional random and dedicated assignment strategy,the proposed methods could optimize RMFS by matching shelves location with Stock Keeping Units(SKU) demand pattern dynamically,which was save about 30% traveling distance and improve the operation efficiency of RMFS.

Key words: robotic mobile fulfillment system, automatic picking system, dynamic storage location reassignment optimization, part-to-picker, online retail

摘要： 针对动态需求背景下的基于移动机器人的拣货系统(RMFS)货架储位再指派问题,提出了考虑货架与储位匹配程度的动态储位再指派策略,并以货架需求频率与储位距离关联性为优化目标,构建了反映仓库布局连续动态变化关系的混合整数规划模型,设计了求解大规模动态货架储位再指派问题的单时刻交换模拟退火(SA-STE)算法。结果表明,SA-STE算法能在合理的时间内寻找到接近CPLEX的近优解,而且在求解大规模问题时具有明显的时间优势。与传统的随机和固定位置指派策略相比,提出的模型和算法可以实现库存存货单元存储结构与需求模式相匹配的优化效果,缩短了30%左右的拣货距离,大幅度提高了RMFS的拣货效率。

关键词: 基于移动机器人的拣货系统, 自动拣货系统, 动态货位指派优化, 货到人, 电商