A knowledge-based two-population optimization algorithm for distributed heterogeneous assembly permutation flowshop scheduling with batch delivery and setup times

Distributed heterogeneous assembly factory environment has become the mainstream of manufacturing enterprises in the real world. Constraints such as batch delivery and setup time are often involved in this distributed heterogeneous assembly environment. To address the problems of low solution quality and slow convergence due to the coupling of these constraints, this paper proposes a Knowledge-Based Two-Population Optimization algorithm (KBTPO) to solve the Distributed Heterogeneous Assembly Permutation Flowshop Scheduling Problem with Batch Delivery and Setup Time (DHAPFSP-BD-ST) with the objective of minimizing inventory and tardiness costs. The algorithm adopts a memetic algorithm as the primary search framework, with reinforcement learning algorithm assisting in collaborative search. The strategy based on knowledge representation and transfer is used to strengthen the communication between populations to accelerate the convergence and improve the efficiency of the algorithm. In addition, several local search operators specific to the problem are designed to enhance the development ability of the algorithm. Comparative experiments show that KBTPO is superior to advanced algorithms in convergence speed and quality. This algorithm is very suitable to solve the distributed heterogeneous scheduling scenarios in the real world, and has9 important significance for the actual manufacturing scheduling optimization.