Optimization of Multi-Product Aggregate Production Planning Using Improved Genetic Algorithm

Abstract

Medium-term production planning with aggregate production planning (APP) is a crucial step in the manufacturing industry's supply chain. The essential phase determines the production size of each product over a planning horizon. Poor planning will undoubtedly directly impact the company regarding production costs and profits. The aggregate production planning is classified as NP-Hard combinatorial problem. Thus, a powerful approach is required. Most models in aggregate production planning consider a single product. This study modeled aggregate production planning to address a multi-period and multi-product. Thus, a more complex mathematical model is required. Implementing genetic algorithms (GA) may solve the problem with reasonably good solutions. This study aims to improve the GA by applying real-coded chromosomes and the adaptive change of crossover and mutation rates based on predetermined change criteria. The planning produced by the modified genetic algorithm is compared to the manufacturer's actual planning to prove the proposed approach's effectiveness. A set of computational experiments proves that adaptive evolution enables the genetic algorithm to balance its exploration and exploitation ability and obtain better solutions. The modified GA produces a less fluctuating pattern of the production amount. Even though the modified GA yields more inventory cost, the high cost of recruiting new workers can be eliminated. Using the proposed approach, the company can reduce 9 percent of the production cost.