Optimization in job shop scheduling problem using Genetic Algorithm (study case in furniture industry)

Abstract

Job shop scheduling problem belongs to a class of NP-Hard problems. We solve a scheduling problem in a job shop based furniture company. The company produces several products such as chair, table, home decorations, and home accessories. Currently, the company schedules the order using Earliest Due Date (EDD) and First Come First Serve (FCFS) methods. The best schedule resulted from those methods is then chosen and used as the initial solution for Genetic Algorithm (GA) method. The proposed algorithm is implemented in MATLAB 2019a to minimize the makespan. Parameters used in the GA formation of new generations are done by crossover using the Precedence Preservative Crossover (PPX) method and mutations using job-pair exchange mutations. The selection of chromosomes for regeneration in the crossover process is chosen by two chromosomes that have the best fitness and for the mutation process, one chromosome that has the worst fitness is chosen. Solution from genetic algorithm is better than EDD for the case study. From the results, GA produces shorter makespan compared to EDD and FCFS methods. The EDD method gives a makespan of 104,280 minutes and the FCFS method gives a makespan of 118,440 minutes, while GA provides a makespan of 81,780 minutes.