Co-evolutionary algorithm with MRF-based decomposition mechanism for flexible scheduling with uncertainty

Flexible scheduling problem plays an important role in many real world applications as it can model many practical issues, such the power grid scheduling, the logistics scheduling, etc. Most researchers assume that the parameters used in flexible scheduling problems are fixed. However, the uncertain factors in real world problems cannot be ignored. In this paper, we consider the uncertain processing times which are modeled with the uniform distribution and design a hybrid coevolutionary algorithm with Markov random filed (MRF) based decomposition mechanism for flexible scheduling problem with uncertain processing times. All decision variables are decomposed according to the learned network structure and the parameters of MRF. The proposed MRF-based decomposition mechanism is embedded into a cooperative co-evolutionary algorithm which is named as CEA-MRF. Finally, the simulation results demonstrate the effectiveness of our proposed algorithm compared with the state-of-the-arts.