IMPACT OF GENETIC ALGORITHM OPERATORS ON ITS PERFORMANCE IN SOLVING FLOW SHOP SCHEDULING PROBLEMS

The primary objective of flow shop scheduling is to obtain the best sequence which optimizes various objectives such as makespan, total flow time, total tardiness, or number of tardy jobs, etc. Due to the combinatorial nature of the flow shop problem (FSP) there is a lot of artificial intelligence methods proposed to solve it. The Genetic Algorithm (GA), one of these methods, is considered a valuable search algorithm capable of finding a reasonable solution in a short computational time. GA operators, (selection, crossover and mutation process), give different forms that can be combined to give various GAs. In this paper we investigate the impact of selection, crossover and mutation process on the quality of the GA solution in solving the flow shop scheduling problems. In this study, four selection methods, seventeen crossover methods and eight mutation methods are investigated. The computational results show that there are significant differences among the investigated methods on the performance of the proposed GA.