An estimation distribution algorithm for wave-picking warehouse management

Recently, market has witnessed a tremendous growth in E-commerce sales, which bring tons of opportunities as well as challenges. Warehouses have to handle unique characteristics of customer orders in the era of E-commerce which consists of small order scales, large items count, unexpected irregular order arrival patterns, seasonality demand peeks, and high service level expectations. Warehouses are adopting wave-picking as an effective policy composed of item-batching, load-assignment and picker-routing problems. In this research, principle combination of load-assignment and picker-routing problems is studied. A mixed integer mathematical model is established based on features of a wave-picking warehouse. In order to conquer the complexity caused by routing decision of the proposed problem, a set of effective modified Estimation Distribution Algorithms is developed. The set of proposed algorithms is proved to have stable gaps (1% on average and maximum less than 2%) compared with Cplex 12.8, while can be solved in much larger scale within quite short time (100 pickers and 350 items in each wave within less than 2 min).

Automated summary

The rapid growth of E-commerce sales has brought opportunities and challenges to warehouses, and adopting the Wave-Picking strategy can effectively address these issues. This research establishes a mixed integer mathematical model based on the characteristics of a Wave-Picking warehouse and develops a set of effective algorithms to solve the load-assignment and picker-routing problems.