Research Of Flexible Job Shop Scheduling Problem Based On Artificial Fish Swarm Algorithm

The job-shop scheduling is to meet some of performance indicators by reasonable arranging the production resources, which is the key factor for improving the market competitiveness of manufacturing enterprises. Compared to the classic job-shop scheduling problem (JSP), flexible job-shop scheduling problem (FJSP) increases the flexibility of machines selection, making it closer to the actual production environment. Thereby the research of solving FJSP has important practical application value. In this paper, a latest swarm intelligence optimization algorithm which is artificial fish-swarm algorithm (AFSA) is used to respectively optimize the single-objective FJSP and multi-objective FJSP. The main works are summarized as follows:(1) The flexible job-shop scheduling problem needs to solve machine selection sub-problem and operation scheduling sub-problem. When one problem is solved, it may affect other problem to solve, so two sub-problems are mutual restriction and influence. The pre-principle arranging mechanism and post-principle arranging mechanism are respectively designed to enhance the diversity of population by adjusting machine assignment and operation sequence with different orders.(2) An artificial fish swarm algorithm with estimation of distribution (AFSA-ED) is proposed for the FJSP with the criterion to minimize the makespan. In this algorithm, a preying behavior based on estimation of distribution is introduced to improve the optimizing process of algorithm, and an attracting behavior is proposed to improve the global exploration ability of algorithm. Besides, a critical path search strategy is proposed to enhance the local exploitation ability of algorithm. Simulated experiments are carried on 160 benchmarks and the results are compared with other optimization algorithms to prove the validity of our algorithm.(3) For the three-objective FJSP with makespan, maximum machine loading, and total machine loading, a cooperative hybrid artificial fish algorithm is designed. In searching process, the algorithm realizes global search by the cooperation of fish swarms and local search by cooperating with simulation annealing algorithm. Besides, we propose an improved ε-Pareto dominated strategy for evaluating the fitness, also crowding distance and elitist archive are used to keep the diversity of population. Finally, experiments show that our algorithm can get better quality non-dominated solutions.