Application of genetic algorithm (GA) and design of experiment (DOE) to solve the inventory dynamic lot-sizing problem: a case study

Abstract

In inventory planning, the Dynamic Lot-sizing (DLS) problem is a challenging problem characterized by time-varying demand and quantity-varying costs. Optimization methods using mathematical models have difficulty solving this problem due to its combinatorial nature and dynamic characteristics. Genetic algorithm (GA) is a meta-heuristic algorithm that can search for near-optimal solutions in a large solution space, thereby effectively solving this problem. In this study, the mathematical model of the DLS problem has been constructed. Based on the model, a GA algorithm is developed and used to solve the DLS problem with the objective to minimize the total inventory cost. The parameters of the GA algorithm are optimized by using Design of Experiment (DOE). The results show that GA is better than the currently used heuristic algorithm, and GA with DOE is better than GA.