A competitive memetic algorithm for the distributed flow shop scheduling problem

Distributed scheduling problems have caught more and more attention during recent years. Due to the difficulty and complexity in solving the problems, it is very important to develop effective and efficient solution algorithms. In this paper, a novel optimization method called competitive memetic algorithm (CMA) is proposed. In the CMA, multiple sub-populations are employed to perform search among the solution space with different neighborhoods. Sub-populations compete with each other for generating different number of their own offspring in every generation. Meanwhile, sub-populations share the search information for collaborating with each other. Besides, local search operators are adopted to intensify local exploitation. Then, with the permutation based encoding scheme and the earliest completion factory (ECF) rule as well as some problem-dependent neighborhoods, the CMA is applied to solve the distributed permutation flow shop scheduling problem (DPFSP). To test the performances of the CMA, numerical results based on 720 large-sized instances are presented, and the comparisons to the existing algorithms are provided. The effectiveness of the CMA in solving the DPFSP is demonstrated.