Research And Application Of Optimization Algorithm For Fuzzy Flexible Job Shop Scheduling

Production scheduling is the core and key technologies of the enterprise production management system.However,with the expansion of the factory scale and the complexity of the product processes,the randomness and variability of the production system are increasingly highlighted,which makes the research on scheduling problems become more difficult.Under the background of the vigorous development of fuzzy theory,the fuzzy feature can effectively reduce the decision deviation caused by uncertainty during the schedule,and improve the reliability of the scheduling scheme.In addition,considering the enterprises’ urgent demand of flexible production model,high quality scheduling in the flexible job shops with fuzzy features has become the focus of the industry and academic community.This paper focuses on the single-objective and multi-objective fuzzy flexible job shop scheduling problems,and optimizes them by improving the quantum-behaved particle swarm optimization.The following is the main research content:(1)Using fuzzy numbers to represent production parameters,the modeling method of fuzzy flexible job shop scheduling problems is given,and the classification of models and commonly used objective function expressions are introduced.The basic quantum-behaved particle swarm optimization is elaborated by analyzing its principle of realization,feasibility and optimization framework of solving the fuzzy flexible job shop scheduling problem.(2)Aiming at the lack of optimization ability of existing intelligent algorithms for solving the single-objective fuzzy flexible job shop scheduling problem,a hybrid quantum-behaved particle swarm optimization is proposed.Combined with the greedy strategy,a process based on single-layer encoding is designed and the particle position is mapped.The boundary correction strategy and cooperative update framework are applied to the quantumbehaved particle swarm optimization to improve its search efficiency in the decision space.At the same time,the performance of the optimization is further improved by fusing crossover operators and path relinking technique.Experimental results show that the proposed optimization is more effective in solving fuzzy scheduling examples and the fiber manufacturing workshop instance than the basic quantum-behaved particle swarm optimization and other algorithms in recent literature.(3)Aiming at the disadvantages of strong randomness and high complexity of using the posterior method in solving multi-objective fuzzy flexible job shop scheduling problems,an improved multi-objective quantum-behaved particle swarm optimization is proposed.On the basis of the realization of two-layer coding and mapping rules,the average optimal position and local attractor in the optimization are optimized by weight factors and random numbers with variable Gaussian distribution.And based on the idea of dynamic grouping,the cooperative update framework is further improved,strengthening the in-depth search ability of the optimization.In addition,in order to control the calculation amount of the optimization process and ensure the diversity of the output non-dominated solutions,an external archive based on the congestion entropy is introduced to achieve the high efficiency of the Pareto frontier maintain.Experimental results show that the proposed optimization performs better than the basic quantum-behaved particle swarm optimization with the external archive and advanced algorithms in existing literature in solving complex examples and the actual shop instance.(4)Oriented by actual demand,the proposed theoretical methods are applied to the scheduling of actual factories,and an intelligent scheduling system for the optical fiber manufacturing industry is developed.The design ideas of the system and the implementation methods of functional modules are introduced in detail.