A Production Planning Problem Solved by the Particle Swarm Optimization

Several researchers have referred to the capacitated production lot-sizing allocation problems as NP-Hard. Consequently, it is more difficult to solve the capacitated production allocation problem considering practical characteristics, such as allocation problems among bottleneck machines, photo masks, and products with different re-entrant layers. In this paper, we proposed a novel variation of the particle swarm optimization (PSO) model, which is a binary PSO model with adaptable inertia weight and mutation mechanism. It is converted to be able to solve the model of binary decision variables. Moreover, it improves some weaknesses, including a propensity for obstruction near the optimal solution regions that hardly improve solution quality by fine tuning. In order to compare effectiveness, the traditional PSO, genetic algorithm, and the proposed PSO in this study are compared by the practical production planning problem in the TFT Array process. Based on the results of the experiments, it can be concluded that the proposed PSO is more effective than the other approaches in terms of superiority of solution and required CPU time.