【作者】 叶明；

【导师】 王宁生；

【作者基本信息】 南京航空航天大学 ， 机械电子工程， 2007， 博士

【摘要】 本文以整车制造企业中非常典型的多级混流制造系统为研究背景,按照从单目标到多目标,从静态调度到动态调度,逐步深入的研究思路对多级混流生产调度系统总体框架,多级混流生产调度问题的优化目标、数学模型、解决策略以及优化算法,模糊信息下的调度方案筛选机制等理论问题和关键技术进行了详细而深入的研究,具体内容如下:针对目前国内整车制造企业的实际需求,从生产成本、均衡生产以及设备利用率等方面考虑,构建了多级混流制造系统调度问题的数学模型。提出了一种递进优化策略,即借助不同制造阶段之间的存储区,运用多目标综合优化与有限柔性下的队列顺序二次优化相结合的优化方式求解整车制造企业的多级混流生产调度问题。在问题的解决策略中,首先提出了一种改进的蚁群优化算法来求解以涂装车间油漆更换成本最少、总装车间零件使用速率均匀化为目标的调度问题;然后在二次优化中考虑了过去混流生产线调度问题通常被忽略存储区结构约束,在详细分析存储区结构特点的基础上,提出了一种用于求解有限柔性下队列顺序调整问题的二次优化算法;最后提出了一种路由算法用于求解生产队列通过存储区时的路径选择问题。仿真试验结果说明了本文提出的解决策略和算法能够获得可行和满意的解。针对多级混流制造系统动态调度问题,采用了一种周期式与事件驱动相结合的调度机制。对于紧急订单、加工质量问题等例外事件设计了相应的解决策略,提出了一种基于层次结构搜索域的多层候选集蚁群优化算法来解决存储区结构约束下的多目标调度问题。仿真试验结果说明了本文的动态调度策略与优化算法在求解多级混流制造系统动态调度问题上的可行性和有效性。针对多级混流制造系统的调度方案筛选问题,本文提出了两种解决方案:第一种是F-AHP与信息熵合成法,第二种是辅助目标决策法。F-AHP与信息熵合成法的具体步骤是首先通过三角模糊层次分析法获得调度人员对于各项评价指标的主观权重系数,然后通过熵权法获得调度方案集包含的各项指标的客观权重系数,从而在方案筛选过程中充分体现了主、客观因素。对于第二种解决方案,本文采用混流装配线停线时间作为辅助目标,首先描述了具有扩展作业域的混流装配线停线问题的数学模型,研究了该问题的若干性质,例如采用可扩展作业域后装配线发生停线的充分条件和必要条件等,并在此基础上提出了一种可扩展作业域下的生产线停线时间计算方法。仿真结果说明了算法的可行性,同时也说明了可扩展作业域对于生产线停线时间具有较大影响。以南京名爵汽车有限公司的制造执行系统为支撑环境,在本文理论研究的基础上,设计和开发了多级混流生产线调度原型系统,经过在南汽名爵的试验性运行,初步验证了该原型系统的可行性和有效性。更多还原

【Abstract】 The background of this dissertation is the multi-level mixed model product line in automobile plant. According to the sequence from single objective scheduling to multi-objective scheduling, from static scheduling to dynamic scheduling, a detailed study on the scheduling system framework model, objectives, strategy and algorithm for the multi-level mixed model product line scheduling problem, and method for the multi-objective scheduling decision making problem is carried out. The main contents and achievements of the dissertation are as follows:For the requirements of automobile plants, the mathematics model of the scheduling problem is presented in details under the considering of product cost, JIT and utility ratio of equipment etc. And a step optimized solution, which through the combination of multi-objective optimization and secondary optimization, is designed to solve the problem. At first an improved ant colony optimization (ACO) algorithm with two layers searching space is designed to solve the scheduling problem which minimize the paint purge times in paint shop and keep the constant usage of part in assembly shop. To the problem that the product order couldn’t be executed correctly in some automobile plant due to ignoring the storage constraint, a resequencing algorithm with limited flexibility is designed based on the detailed analysis of kinds of storage structure. At last, a useful route control approach is presented, which can be used to address the product routing problem. Simulation result shows that the proposed strategy and the algorithm are effective and useful.A periodic and event-driven scheduling model is presented to deal with the changing mixed-model manufacture environment. The schedule strategy is designed to deal with the rush order or rework. Based on the two layers searching space ACO algorithm, a hierarchy searching space ACO algorithm is designed to address the multi-objective scheduling problem under structure constraint of multi-level mixed product line. Simulation results show that the proposed model and algorithm are feasible to the problem.To the multi-objective scheduling decision making problem, two solutions be presented, one is fuzzy analytic hierarchy process (F-AHP) and entropy theory method and the other is assistant objective decision making method. The first solution use F-AHP to get the subjective weight from the fuzzy information of decision-maker, and use entropy theory to get the objective weight from the result matrix. To the second solution, the conveyor stoppage time is used as the assistant object to the decision making problem. At first, a mathematics model of conveyor stoppage problem under the concept of extendable region(ER) is presented. Then combined with the ER concept, several useful properties, such as, necessary and sufficient condition that a conveyor stoppage occurs, lower and upper bounds of the objective function are described. And an algorithm for calculating the conveyor stoppage time is designed. Simulation result shows that the algorithm is effective, and the ER has important effect to the conveyor stoppage time.Finally, based on the Build Control System of Nanjing MG automobile plant, a multi level mixed-model product line scheduling prototype system is developed. Through the test run in Nanjing MG automobile plant, it verifies the feasibility of this system.更多还原