Distributed Flexible Job Shop Scheduling Based on an Improved Genetic Algorithm

Abstract

Aiming at the multi-dimensional and highly complex decision-making characteristics of distributed flexible job shop scheduling, and the local convergence and slow convergence speed of the traditional genetic algorithm, an improved genetic algorithm- the dynamic hierarchical genetic algorithm is proposed. The algorithm enhances the global search ability and local optimization ability of the traditional genetic algorithm through a dynamic hierarchical structure, a multi-neighborhood local search strategy, and adaptive adjustment of the crossover and mutation rates. The simulation results show that after 10 independent runs on several improved Kacem (MK) series test cases, the algorithm has a lower minimum average maximum completion time than the traditional genetic algorithm. Especially in the complex cases of MK03 and MK09, not only is the completion time reduced by 24%, but also the number of iterations is reduced. This fully verifies the ability of the algorithm to improve scheduling efficiency and stability in distributed flexible manufacturing workshops. The completion time of MK09 is reduced by about 24%, and the number of iterations is also reduced. This fully verifies the ability of the algorithm to improve the scheduling efficiency and stability in the distributed flexible manufacturing workshop scheduling problem.