Improved invasive weed algorithm based on double layer search framework for flexible batch scheduling problem

Abstract

Aiming at the flexible job shop batch scheduling problem, an improved invasive weed algorithm based on double-layer search framework was proposed for satisfactory batch scheduling results. A three-layer gene coding method integrating batch, process arrangement and processing machine information was proposed. A two-tier search optimization framework was designed, in which the flexible batch was divided through the batch search layer, and then the batch scheduling scheme was obtained by the iterative optimization of the sorting search layer. In the batch search layer, the solution space of batches was reduced based on the average working hours of processes, and a random number segmentation method was proposed to generate batch schemes; in the sorting search layer, the invasive weed algorithm was used to realize iterative optimization. At the same time, the hierarchical initialization method, hybrid machine selection strategy and three local search operators were designed to improve the search ability of the algorithm and avoid falling into local optimization. Taking the maximum completion time as the evaluation index, the superiority and feasibility of the proposed algorithm were verified from three dimensions: performance experimental analysis, framework experimental verification and example experimental verification.