An Improved Genetic Algorithm for Locations Allocation Optimization Problem of Automated Warehouse

Abstract

In an automated warehouse, class-based storage is a method for storage and retrieval. Class-based storage and storage locations assignment implementation decisions have significant impact oil the required storage space and product picking efficiency. To solve the problem of storage/retrieval frequently and dynamic change storage locations, a mufti-objective mathematical model was formulated for storage locations assignment of the fixed rack system. The rack stability and order picking frequency were incorporated based on the class strategy. Because all objectives in the model are conflicting and the sole optimum solution does not exist, an improved genetic algorithm with pareto optimization and niche technology is developed. The approach adds pareto solution sets and niche technology besides traditional operators. It can search the optimum solution sets that distribute uniformly. The approach ensures storage location assignment optimization and offers an optimization decision making scheme for AS/RS. Computational experience with randomly generated data, sets and in industrial case shows that the policies are more effective than class-based storage policy only, and enhance the operational efficiency of an automated storage/retrieval system, as well as a CIMS system. The improved genetic algorithm can be applied to handle large real life problems efficiently.