A Bayesian-Based Co-Cooperative Particle Swarm Optimization for Flexible Manufacturing System Under Stochastic Environment

In recent years, flexible scheduling attracted considerable attention, motivated by both important practical issues and interesting research problems, especially under stochastic environment. In this paper we discuss a stochastic scheduling problem whose scale arranges from small to large. The objective in the schedule is to minimize the processing time over all of the jobs. However, stochastic environment will bring more difficulty especially for large scale because the sudden change of processing time for each operation may break the current optimal solution and lose efficiency. So, we propose a Bayesian network based particle swarm optimization (BNPSO) for solving this stochastic scheduling problem. Firstly, we use new framework named co-cooperative (CC) evolutionary framework which decompose all decision variables into several small group containing part of decision variables to overcome large scale problems. And then, BNPSO adjust the group scene according to their interactive relationships based on Bayesian network structure. Meanwhile, importing self-adaptive mechanism for parameters in order to satisfy the stochastic environment. Some practical test instances will demonstrate the effectiveness and efficiency of the proposed algorithm.