Solving bi-objective integrated scheduling problem of production, inventory and distribution using a modified NSGA-II

Previous research on the integrated scheduling problems generally considers the production and distribution stages, but ignores the arrangement of on-hand inventory saved in the warehouse. However, the overall optimal solution to meet customer demands must require a high degree of coordination between the production, inventory and distribution stages. Encouraged by this, we study a new production-inventory-distribution integrated scheduling problem, in which the production plan in a homogeneous flowshop environment, allocation plan of on-hand inventory and distribution decision of three-party logistics provider, are considered simultaneously. A mixed integer linear programming model with the goal of minimizing total earliness/tardiness penalty costs and total energy consumption is built for the problem. A modified NSGA-II (abbreviated as MNSGA) is proposed to solve the problem. In the MNSGA, a well-tailored three-layer encoding method is presented for the chromosome representation. The adaptive crossover and mutation operators are developed for a good global search. In addition, an objective-oriented local search operator is designed to improve the local exploitation ability. The superiority of the proposed MNSGA is verified by comparing four popular multi-objective optimization algorithms based on 540 benchmark instances. Furthermore, the proposed integration mode is demonstrated to be effective by comparing it with several sequential scheduling methods commonly used in actual production.

Automated summary

Previous research has overlooked the arrangement of on-hand inventory saved in the warehouse in the integrated scheduling problems. In this study, we propose a new production-inventory-distribution integrated scheduling problem that considers the production plan, allocation plan of on-hand inventory, and distribution decision simultaneously. We develop a mixed integer linear programming model and a modified NSGA-II algorithm to solve the problem. The proposed integration mode is shown to be effective compared with sequential scheduling methods commonly used in actual production.