A bi-level optimization of storage allocation and shelf sequencing in a goods-to-person picking system

With the rapid development of e-commerce economy, Robotic Mobile Fulfillment Systems based on "goods-to-person" warehouse order picking mode are widely used in large e-commerce platforms such as Amazon and Alibaba. A Bi-level optimization algorithm is proposed to jointly optimize storage allocation and shelf sorting for the shelf scheduling problem with multiple storage locations in the "goods-to-person" picking system. The upper-layer optimization uses an improved genetic algorithm to allocate storage locations for the shelves to be picked, adopts a problem-specific knowledge strategy to generate an elite initial population, and introduces crossover and neighborhood search operators with adaptive transformation probabilities to enhance the local search and global optimization capabilities. According to the given storage location, shelf sorting optimization is carried out at the lower level, which is abstracted into the variant traveling salesman problem for modeling and analysis, based on which two neighborhood structures are designed and solved with the variable neighborhood search algorithm. The experimental results show that the performance of the improved genetic algorithm is improved by 2.31% compared with the ordinary genetic algorithm, the combined algorithm has certain advantages over the optimization solver in terms of speed and solution quality, which is of great significance to improve the operational efficiency of the "goods-to-person" picking system.