【作者】 周刚；

【导师】 王凌；

【作者基本信息】 清华大学 ， 控制科学与工程， 2012， 硕士

【摘要】 作为一类典型的复杂生产调度问题，柔性调度具有很强的应用背景且求解难度很大。作为一种新兴的智能算法，人工蜂群算法（ABC）以其机制简单、参数少等特点得到了广泛关注，但在柔性调度问题上的研究相对很少。本学位论文针对两类典型的柔性调度问题，即柔性作业车间调度问题（FJSP）和柔性流水车间调度问题（FFSP），开展有效的ABC算法研究，并编制相应的仿真软件。在详细综述柔性调度和ABC算法研究的基础上，论文主要完成了如下工作：（1）针对FJSP问题，以最小化makespan为目标，提出了一种结合关键路径搜索的ABC算法，通过在跟随蜂阶段基于关键路径的局部搜索操作以及在侦察蜂阶段基于随机搜索的种群更新机制，增强了局部改良和全局探索能力。（2）针对多目标FJSP问题，采用Pareto档案集保存非支配解并参与跟随蜂阶段的交叉操作，提出了一种基于非支配排序和拥挤距离的种群划分机制，进而设计了一种基于Pareto支配的ABC算法。（3）针对模糊FJSP问题，基于模糊数计算调度性能指标和比较个体的性能，在跟随蜂阶段设计基于全局最优解的交叉操作，在侦察蜂阶段设计基于全局最优解的自适应种群调整策略，并使用变邻域搜索机制对最优解进行深入搜索，进而提出了一种基于全局最优解的ABC算法。（4）针对不相关并行机FFSP问题，结合问题特性采用多种邻域搜索丰富搜索行为，进而提出了一种基于排列编码的ABC算法。（5）以纺织企业生产为背景，针对FFSP问题，设计了基于ABC算法的调度仿真演示软件。论文通过大量数值仿真以及与现有代表性算法的比较验证了所提算法的有效性，并通过试验设计方法探讨了参数设置的影响。更多还原

【Abstract】 As a class of typically complex scheduling problem, flexible schedulingproblem has a strong application background while the solving complexity isvery large. As a novel intelligent algorithm, artificial bee colony (ABC)algorithm has received intensive interest due to its simple mechanism and fewparameters, while its research on the flexible scheduling problem is relativelyrare. Focusing on two typical flexible scheduling problems, i.e., flexible jobshop scheduling problem (FJSP) and flexible flow shop scheduling problem(FFSP), this dissertation carries out research on the design of ABC algorithmsand designs the corresponding simulation software. In addition to the detailedsurvey on flexible scheduling and ABC algorithms, the main research work ofthis dissertation is summarized as follows:(1) Aiming at FJSP with the criterion to minimize makespan, we proposean artificial bee colony algorithm combining with the local search based oncritical path. Via the local search based on critical path at the onlooker beephase and the updating mechanism of population with random search at thescout bee phase, local exploitation and global exploration capalities areenhanced.(2) Aiming at the multi-objective FJSP, we use a Pareto archive set tosave the non-dominated solutions and to participate in crossover at theonlooker bee phase, and design a sorting strategy for population based onnon-dominated sorting and crowding distance, and propose a Pareto-basedartificial bee colony algorithm.(3) Aiming at the fuzzy FSJP, we use fuzzy number to calculate scheduleobjective and to compare solutions, and design a crossover operator based onthe Best-so-far solution at the onlooker bee phase, and design a populationupdating mechanism at the scout bee phase, and use the local search strategybased on variable neighbouring search, and finally propose a Best-so-far artificial bee colony (B-ABC) algorithm.(4) Aiming at the hybrid flow-shop scheduling problem with unrelatedparallel machine (HFSP-UPM), we apply multiple local searches according tothe characteristics of the problem to enrich the searching behaviour, anddevelop an artificial bee colony algorithm based on permutation encoding.(5) With the background of the textile manufacturing, we design thecorresponding simulation demo software based on ABC algorithm for FFSP.A large number of computational tests and comparisons to the typicalexisting algorithms demonstrate the effectiveness of the proposed algorithms.We also investigate the effect of parameter setting on the algorithms by usingdesign of experiment.更多还原