An improved particle swarm optimization for large-scale permutation flowshop scheduling

An improved PSO for large-scale permutation flowshop scheduling problem (PFSP) is proposed in this paper. Firstly, an adaptive rejection computing strategy is proposed for PSO to avoid it trapping into local optimum early. Secondly, an adaptive nonlinear changing cognition factor and social factor is developed for a better balance between the global exploration and local searching of the mixed particles. Thirdly, according to the discreteness of PFSP, a position sequence method (PSM) for the transformation from continuous problem space to the discrete problem space is proposed to effectively reduce computing time. Fourthly, according to the reversibility of makespan of PFSP, we present a fast computation approach for makespan evaluation. Finally, a large number of simulation experiments are performed on the classical Taillard benchmark. The results show great effectiveness of our proposed algorithm.