A decomposition based multiobjective genetic algorithm with adaptive multipopulation strategy for flowshop scheduling problem

Recently, the solution algorithm for multiobjective scheduling problems has gained more and more concerns from the community of operational research since many real-world scheduling problems usually involve multiple objectives. In this paper, a new evolutionary multiobjective optimization (EMO) algorithm, which is termed as decomposition based multiobjective genetic algorithm with adaptive multipopulation strategy (DMOGA-AMP), is proposed to addressmultiobjective permutation flowshop scheduling problems (PFSPs). In the proposed DMOGA-AMP algorithm, a subproblem decomposition scheme is employed to decompose a multiobjective PFSP into a number of scalar optimization subproblems and then introduce the decomposed subproblems into the running course of algorithm in an adaptive fashion, while a subpopulation construction method is employed to construct multiple subpopulations adaptively to optimize their corresponding subproblems in parallel. In addition, several special strategies on genetic operations, i.e., selection, crossover, mutation and elitism, are also designed to improve the performance of DMOGA-AMP for the investigated problem. Based on a set of test instances of multiobjective PFSP, experiments are carried out to investigate the performance of DMOGA-AMP in comparison with several state-of-the-art EMO algorithms. The experimental results show the better performance of the proposed DMOGA-AMP algorithm in multiobjective flowshop scheduling.