【作者】 桂玲；

【导师】 谢勇；

【作者基本信息】 华中科技大学 ， 系统工程， 2017， 硕士

【摘要】 目前JIT(just in time)生产模式被越来越多地应用于实际的建造和生产环境中,单机的提前/拖期调度问题也被更多学者重视。本论文为了对提前/拖期调度问题进行细致而深入的研究,主要从三个方面对该问题进行了研究:单机的提前/拖期调度问题;单个时间窗的提前/拖期问题;所有交货期均为时间窗的提前/拖期问题。本论文首先研究了最基础的情形——单机的提前/拖期调度问题。本文在最优开工时间算法的基础上提出了劣序剔除规则,该剔除规则解决了任意工序的调度问题,对劣序列进行了筛选和剔除,提高了问题的收敛性。此外本文还提出了改进的遗传算法,将最优开工时间算法以及劣序剔除规则与遗传算法相结合,运用改进的遗传算法对港珠澳大桥建造过程中的实际需求计划进行了调度,并且通过对遗传算法和改进的遗传算法进行多组对比实验证明了改进的遗传算法在成本和收敛速度上都有所改进,进而证明了劣序剔除规则的有效性。然后本论文对单个时间窗的提前/拖期调度问题进行了研究。对于该问题,本文从两个方向进行了研究:分析时间窗对总成本和最优开工时间的影响;对于单个工件的交货期为时间窗的问题进行建模分析。通过分析对时间窗的范围以及时间窗对于调度方案的开工时间以及成本的影响进行了讨论,并得出了相应的结论。最后对于所有工件的交货期均为时间窗的问题,本文运用遗传算法与禁忌搜索算法结合的混合策略进行求解。通过对遗传算法和混合策略的多组对比实验,我发现混合策略在迭代次数与总成本上都明显优于遗传算法,并且随着工件数量的增加,优势愈加明显,证明了混合策略的有效性。总而言之,本文对单机器的提前/拖期调度问题进行了全面而深入的研究,但是后续仍有许多问题亟待解决和研究。更多还原

【Abstract】 With the JIT(just in time)production model is more and more applied to the actual production and construction environment,single machine scheduling problem with early / tardy penalties has been getting more and more attention by scholars.In order to study the problem of early / tardy scheduling,this paper mainly studies the problem from three aspects: the early / delay scheduling problem of single machine;early/tardy scheduling problem with single due window;early/tardy scheduling problem with distinct due windows.First of all,this paper studied the basic situation-the scheduling problem with early / tardy penalties.Based on the optimal start time algorithm,this paper proposes a inferior sequence filtering rule,which solves the scheduling problem of any process,filters and eliminates the bad sequence,and improves the convergence of the problem.In addition,this paper proposes an improved genetic algorithm,which combines the optimal start time algorithm and the inferior sequence filtering rule and the genetic algorithm.The improved genetic algorithm is used to schedule the actual demand plan in the construction of the Hong Kong-Zhuhai-Macao Bridge.And through the comparative experiments between genetic algorithm and improved genetic algorithm,it is proved that the improved genetic algorithm has improved the cost and convergence speed,which proves the validity of the inferior sequence filtering rule.Then the paper studies the early/tardy scheduling problem with single due window.For the scheduling problem,the paper studied the problem from two sided:analyze the impact of the due window on the original scheduling scheme;analyzes the problem by modeling.By analyzing the scope of the due window and the influence of the due window on the time and cost of the scheduling scheme,we have got some important conclusions.For the scheduling problem with distinct due windows,this paper uses a hybrid strategy of genetic algorithm and tabu search algorithm to solve the problem.Through the comparison of the hybrid strategy and genetic algotithm,we find that the hybrid strategy is much better than the genetic algorithm both in the number of iterations and total cost.And with the increase of the number of jobs,the advantage becomes more and more obvious,which proved the effectiveness of the hybrid strategy.All in all,this paper has conducted a comprehensive and in-depth study on the single-machine scheduling problem with early/tardy penalties,but there are still many problems that need to be solved and studied.更多还原