A hybrid genetic algorithm and tabu search for minimizing makespan in flow shop scheduling problem

Abstract

This paper combines the tabu search process with a genetic algorithm by employing a new partial opposed-based as the population initialization technique to minimize makespan. Flow shop is a prominent variety in scheduling with various applications in numerous disciplines, including production. Its difficulty makes many researchers develop algorithms to solve it since it is categorized as an NP-hard problem. Some approaches have been developed to resolve production scheduling, especially evolutionary algorithms. The genetic algorithm becomes the most used algorithm in evolutionary to address production schedule because of its capability to produce good, fast, and efficient results to explore complex solution space (global search). However, it provides ineffective results when doing exploitation (local search) that is easily stuck in the optimum local area. Contrarily, tabu search has superiority in local search to help genetic algorithms not be trapped in a local optimum. By combining these two algorithms with the initialization scheme, a new algorithm that balances searches in resulting higher quality solutions is obtained. The proposed algorithm was tested using 120 problem instances to carry out the algorithm performance. The empirical results state that the developed algorithm can improve 115 solutions out of 120 instances than the six other hybrid algorithms.