A novel topological neighborhood structure for flexible job shop scheduling problem with variable sublots

Lot streaming is an effective approach to reduce machine idle time by splitting each operation into sublots, enabling parallel processing. However, the search space will also expand dramatically due to the variability of batching and scheduling. Neighborhood structure is an effective approach to obtain high-quality solutions with less computational effort in a complex search space. This paper proposes a novel neighborhood structure, and uses it to develop an effective algorithm for flexible job shop scheduling problem with variable sublots (FJSP-VS). Firstly, a three-dimensional disjunctive graph is developed to represent solutions clearly by incorporating an axis of batching. This representation captures comprehensive neighborhood features, and provides a robust basis for neighborhood perturbations. Subsequently, a novel topological neighborhood structure is proposed for deeper exploration, which can effectively avoid the generation of infeasible solutions while ensuring the quality of neighborhood solutions. The novel topological neighborhood structure is integrated into a variable neighborhood search component, for effective local searching. On this basis, a topology-guided memetic algorithm (TGMA) is proposed, which can obtain high-quality solutions in the complex solution space. Experiments are organized on expanded benchmarks of varying scales, and the proposed TGMA can obtain better solutions than several state-of-the-art algorithms in over 90% instances. The results demonstrate its superior performance in solution quality and computational efficiency when solving the complex high-dimensional FJSP-VS.