基于差分进化算法的柔性作业车间调度问题研究

发布时间：2018-04-30 08:26

本文选题：柔性作业车间 + 差分进化算法　；参考：《华中科技大学》2014年硕士论文

【摘要】：柔性作业车间调度是在实际制造生产中广泛存在的一类问题。对该问题的研究，可以有效提高车间的生产效率，缩短制造周期。此外，实际生产调度问题还具有多目标、动态性等特点，同时需要对生产中各种突发事件进行及时的响应。本文首先研究了经典的柔性作业车间静态调度问题。在该问题的研究中，提出了一种“预调度确定各工序的加工机器”的优化策略，并将其应用到差分进化算法的种群初始化中，提高初始种群的质量。同时，提出了一种新的种群改进策略，，在算法进化陷入局部最优解，最优解一段时间不改进的情况下，适时得去改进种群的质量。将该策略融合到差分进化算法的框架中得到改进差分进化算法。通过和其他算法的比较验证本文提出的改进差分进化算法求解性能优越。随后，本文研究了不同再调度周期下的柔性作业车间动态调度问题。通过模拟随机工件到达的生产环境，运用周期性再调度的调度策略将各个工件依次划入到对应的调度区间去进行求解。在各个调度区间上，以效率和稳定性为目标，设计一种基于Pareto概念的多目标差分进化算法对该调度区间的工件进行调度优化，并提出了一种二级选择策略应用于多目标算法中，最后从优化算法获得的非支配解集中采用决策策略选出一个调度方案作为实际调度加工方案。通过研究在不同的再调度周期下，对先后到达相同数量的工件进行调度得到的最后的完工时间、总拖期、总效率和总稳定性之间的差异，对结果进行分析，得出了不同再调度周期对各个性能指标的影响，便于指导生产实践。最后，本文研究了不同动态事件下的柔性作业车间动态调度问题。考虑了机器故障/修复，紧急订单到达，普通订单到达等动态事件，采用基于周期与事件驱动的再调度策略。在窗口工件的调度优化中，以完工时间，总拖期，总偏离度为优化目标，并设计了Pareto决策策略从最后的非支配解集中选择出一个合适的方案作为实际调度方案。通过实例测试，比较了在动态事件发生时，再调度前后调度方案的变化。
[Abstract]:Flexible job shop scheduling is a widespread problem in actual manufacturing. The research on this problem can effectively improve the production efficiency and shorten the manufacturing cycle. In addition, the actual production scheduling problem also has the characteristics of multi-objective, dynamic and so on, and needs timely response to all kinds of unexpected events in production. In this paper, the classical static scheduling problem of flexible job shop is studied. In order to improve the quality of the initial population, an optimization strategy of "pre-scheduling and determining the processing machines in each process" is proposed in this paper, and it is applied to the population initialization of the differential evolution algorithm (DEA) to improve the quality of the initial population. At the same time, a new population improvement strategy is proposed to improve the population quality in time when the algorithm evolves into a local optimal solution and the optimal solution is not improved for a period of time. The strategy is fused into the framework of differential evolution algorithm (DEA) and improved differential evolutionary algorithm (DEA). Compared with other algorithms, the improved differential evolution algorithm proposed in this paper is superior to other algorithms. Then, the dynamic scheduling problem of flexible job shop under different rescheduling periods is studied. By simulating the production environment arrived by random jobs, the scheduling strategy of periodic rescheduling is used to assign each job to the corresponding scheduling interval in turn to solve the problem. Aiming at efficiency and stability, a multi-objective differential evolutionary algorithm based on Pareto concept is designed to optimize the scheduling of jobs in each scheduling interval, and a two-level selection strategy is proposed to apply to the multi-objective algorithm. Finally, a scheduling scheme is selected from the non-dominated solution set obtained by the optimization algorithm as an actual scheduling scheme. By studying the difference between the final completion time, the total delay period, the total efficiency and the total stability obtained by scheduling the same number of jobs in different rescheduling cycles, the results are analyzed. The influence of different rescheduling cycle on each performance index is obtained, which is convenient to guide production practice. Finally, the dynamic scheduling problem of flexible job shop under different dynamic events is studied. Dynamic events such as machine fault / repair emergency order arrival and general order arrival are considered and rescheduling strategy based on periodicity and event driven is adopted. In the scheduling optimization of the window workpiece, the completion time, total delay time and total deviation are taken as the optimization objectives, and the Pareto decision strategy is designed to select a suitable scheme from the final non-dominated solution set as the actual scheduling scheme. The change of scheduling scheme before and after rescheduling is compared by example test.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TH186;TP18

【参考文献】