【作者】 周伟；

【导师】 邵新宇； 饶运清；

【作者基本信息】 华中科技大学 ， 工业工程， 2010， 博士

【摘要】 本文根据舱体类产品的生产特点,对其生产过程中出现的分批调度问题进行了研究,并重点研究了两类问题,即有限能力单级生产系统分批调度问题和有限能力多级生产系统分批调度问题。通过对上述两类问题的描述与求解策略分析,逐步建立了舱体类产品分批调度优化及其执行过程管理的总体研究框架。舱体类产品关键零部件分批调度属于有限能力单级分批调度问题。与一般的单级问题相比,该问题具有多产品、多机、有限能力及定批量等多个约束的特征。针对该问题,本文利用遗传算法,构造了一个能够同时考虑产品数、设备数及生产周期的复合基因,并设计了产品交叉和逐层两点交叉相结合的交叉方法,最后运用全空间求解策略,得到了能够同时满足所有约束的解。并通过实例分析,证明了该方法的有效性。舱体类产品装配生产分批调度是典型的有限能力多级分批调度问题。该问题除具备单级问题的约束外,还需要考虑部件之间的先后约束关系,增加了问题求解的复杂性。本问题以生产成本最低为目标,建立相应的数学模型,并提出了一种基于规则的启发式算法进行求解。该算法是一种两层迭代算法,它通过在上层利用遗传算法进行分批计划计算,下层基于规则排序对上层分批方案进行验证的方式来迭代求解,对实例计算结果的分析验证了算法的有效性。以上述理论研究为基础,进行了执行过程管理研究。针对生产进度控制、物料控制和质量控制等生产过程中存在的问题,分别提出了相应的过程管理方法。通过该方法,使得生产进度数据、物料信息、质量数据能够被有效的采集及分析,实现了对生产过程的实时监控,保障了分批调度计划的正确执行。针对扰动环境下的执行过程管理问题,文中又提出了一种同时考虑生产总成本与新旧方案差异成本的双目标模型,并通过数值实验分析,验证了模型的有效性。以某舱体类产品生产企业为应用对象,在分析其现状的基础上,以上述理论方法为基础,设计了一套能同时支持单级与多级分批调度问题的制造执行系统。通过该系统在相关企业的成功应用,验证了本文所研究的分批调度优化算法与执行过程管理方法的可行性与实用性。最后,对全文所做的主要工作进行了总结,并对后续的研究工作进行了展望。更多还原

【Abstract】 This dissertation aims at the lot sizing and scheduling problems (LSP) which are always presented within the production of the Cabin products, and two kinds of LSPs which are the capacitated single-stage LSP and the capacitated multi-stage LSP are primarily researched in the thesis. Through the description and solution strategies analysis of the above two problems, an overall research framework of the lot sizing and scheduling optimization and the process management of the Cabin product is established.The LSP of key components of Cabin product belongs to the capacitated single-stage LSP. Compared with the common single-stage LSP, this problem has more constraints, such as multi-product production, multi-machine, capacitated and lot-size consideration. A complex gene with the information of the product numbers, machines and the periods is proposed. The full-space solution strategy is used to get the solution and it can satisfy the whole constraints. Then the validity of the method is proved by a case analysis.The LSP of the Cabin assembly production is a classic multi-stage LSP. Beside the constraint of single-stage LSP, the component precedence which is a specail constraint makes the problem more complex. Considering the objective function of minimum production cost, a mathematical model is established and solved by a rule-based heuristic method with two iterative layers. The above layer deals with the lot sizing calculation and the second schedules the jobs according to the above results. The validity of the algorithm is proved by analysis of the case data.Based on the above-mentioned theory, the process management is studied. For the production process control, material control and quality control in the process management situation, some methods are proposed. Then the process data, material information and quality data are collected and analyzed through these methods. As a result, the real-time monitoring of the production is realized and the plan is executed correctly. To facing the problem under a turbulence environment, a Bi-objective model is proposed taking into account the products cost and the new-old difference cost. Then the validity of this model is proved by a case analysis. Taking the example of Cabin product enterprise, a Manufacturing Execution System is developed to support the single-stage and multi-stage LSP simultaneously. The feasibility and practicability of the methods mentioned in this dissertation are proved through the successful implement in this enterprise.At last the research work is concluded and the future research efforts are proposed.更多还原