【作者】 宋潇潇；

【导师】 孙棣华；

【作者基本信息】 重庆大学 ， 控制理论与控制工程， 2010， 博士

【摘要】 生产调度是生产管理中的一个核心问题,其实质是合理地利用有限的资源来完成生产任务分配,以达到一个或多个性能指标的组合优化问题。研究生产调度问题对实施调度的具体生产对象具有很强的依赖性,必须针对性地考虑生产背景。同时,调度过程的指导思想以及调度相关信息的获取都是调度得以实现的基本前提。流水线调度(Flow shop)是受到广泛关注的研究问题,基于传送带的流水线生产是制造行业的一种典型生产模式。与传统的流水线作业方式不同,传送带式流水线的线上工件加工速度始终统一,加工时间不仅与工件自身有关,还受到线上其他工件加工时间的影响。由于其特殊的物理结构以及生产过程中的大量约束,加之现代生产条件下众多的不确定因素,导致其较一般的流水线调度问题更为复杂。目前,针对此类流水线的研究特别是关于生产调度的研究相对较少,阻碍了生产系统性能的改善和效率的提高。另一方面,精益生产是一种面向多品种、小批量生产的现代生产管理思想,以其为指导研究生产调度问题,在调度的目标、信息需求以及约束条件等方面都将产生巨大变化。本文以传送带式流水线为研究对象,引入精益生产管理思想研究生产调度问题。首先研究理想生产信息环境下的传送带式流水线的调度问题,然后重点从实际调度过程中生产信息获取的不确定性,以及生产过程中出现的不确定因素这两个方面来研究生产调度问题。针对生产信息获取的不确定性问题,根据实际调度过程中对生产信息的需求,研究了RFID与条码相结合的自动标识技术,为实施精益生产和传送带式流水线生产调度提供生产信息保障;在该标识技术基础之上,讨论了其信息采集的不稳定性对生产调度的影响,进而研究了生产信息非理想情况下的传送带式流水线调度模型。针对生产过程中出现的不确定性因素,考虑到生产过程中由于设备故障或订单变动等因素带来的影响,分别从静态和动态两个角度对传送带式流水线上的重调度问题进行了研究。最后,针对现有算法在求解此类调度问题时编码和迭代方式方面存在的缺陷,研究了基于极坐标概率编码的改进人工鱼群算法。论文主要内容包括:①分析了面向精益生产的传送带式流水线生产调度的特点,提出理想条件下的传送带式流水线生产调度模型。研究传送带式流水线的物理结构和生产模式,分析精益生产中准时制和“零等待”等思想对调度模型的影响,指出现有调度模型不能准确描述此类流水线的实际情况。提出了面向精益生产方式的传送带式流水线调度模型,考虑流水线中线上工件加工速度统一、不同工件之间存在调整间隔和一条流水线上同类工件连续加工等特点,引入并行调度问题,建立了并行传送带式流水线调度模型。②针对实际生产调度对生产信息的需求,以传送带式流水线为研究对象,提出结合RFID与条码技术的对象标识与信息采集技术。分析生产信息对实施面向精益生产的传送带式流水线生产调度的重要性,提出一种基于RFID与条码相结合的对象标识与信息采集框架,按照种类和信息支持程度的需求,将生产过程中的零部件和工件按照不同的标识粒度进行管理;从物理层和数据层两个方面进行研究,建立一种RFID与条码结合的对象标识和信息采集技术,为保证生产信息的可靠性和准确性提供支持。③在标识技术应用的背景下,研究考虑生产信息不稳定情况下的生产调度问题,提出一种信息非理想情况下,面向精益生产的传送带式流水线调度模型,并根据并行调度思想,建立并行调度模型。针对RFID与条码相结合的对象标识与信息采集技术在传送带式流水线上应用过程中,出现的信息采集时间偏移和漏读的情况,结合传送带运行速度的限制,提出一种考虑信息非理想情况下的单流水线生产调度模型,以及并行流水线调度模型。④在求解传送带式流水线调度问题时,针对传统算法在求解过程中存在的缺陷,提出一种基于极坐标编码的改进人工鱼群算法。针对传统算法在求解传送带式流水线生产调度问题时,编码和迭代方式方面存在的缺陷,研究一种基于极坐标编码的改进人工鱼群算法。借鉴概率编码思想,提出一种基于极坐标的概率编码方式;在此基础上,对人工鱼群算法的觅食、聚群和追尾行为进行改进,并根据生产调度求解过程中的特点提出一种格雷码译码方式;最后分析了该算法在求解调度问题中的优势,并通过实验验证算法的有效性和稳定性。⑤使用基于极坐标的改进人工鱼群算法对调度模型进行求解。详细分析求解过程中所用到的编码方式、迭代方式、人工鱼个体距离、最优保留机制和约束违反度策略等关键技术;对不同参数的算法求解效果进行比较,选取最优参数的改进算法与人工鱼群算法进行对比实验,对生产调度方案进行优化。实验验证表明:使用理想情况下的调度模型描述生产过程会产生明显偏差,信息非理想情况下建立的模型则更能准确地描述传送带式流水线上实施面向精益生产的生产调度。⑥根据调度方案执行过程中出现的设备故障或订单变动等带来的不确定性因素,研究了传送带式流水线上实施重调度的方法。分别从静态和动态两个方面分析传送带式流水线上的重调度问题,针对并行流水线中的设备故障和订单变化的问题进行研究,提出了进行重调度的方法;通过比较实验验证了提出的方法在此类流水线上进行重调度的有效性。上述研究工作及结果为在传送带式流水线上实施有效的生产调度和精益生产管理提供了理论和实际应用方面的支持,丰富了流水线调度问题的研究内容、技术和方法,并为制造企业实施精益生产,改善生产调度水平,进而提高生产效率及减低制造成本提供了一条有效的途径。更多还原

【Abstract】 Production scheduling is one of the key problems in this requirement. Production scheduling is a problem, which concerns the limitied resources to complete tasks in order to optimize one or more objectives. The implement objective, scheduling ideology and information gethering are the key problems in production scheduling.Scheduling in flow shop is one of the most popular research problems, and the production mode with conveyor belts is a typical mode in manufacturing industry. Different from other flow shop, the jobs have the same operation speed at the same time on line. The operation time of any jobs not only decided by itself, but also by other jobs on line. For its special structure, constrains in procrssing and uncertaint elements, the researches on this line especially on the scheduling problems are few. It hinders the improvement of production efficiency in this kind of line. Lean Production (LP) is management thought which is created facing small-batch and lot-variety production. It changes the conditions, information requirements and scheduling objectives when introduce LP into scheduling problem.Making the conveyor belt flow shop as the study object, introducing LP management thought, scheduling with LP in conveyor belt flow shop is researched. After study the scheduling problem in conveyor belt flow shop, a scheduling model, in ideal production information environment, is proposed. Then, the research foucs on the real scheduling based on unstable information gathering and the emergencies in processing. Aiming at the information requirements in scheduling, an automatic identification technology, combining the advantages of RFID and barcode technology, is proposed to ensure the information support for LP. Based on the application situation of the identification technology, a kind of scheduling model of conveyor belt flow shop with unstable information is proposed. Then, Considering the effects of machine failures and rush orders in scheduling, research the rescheduling problems in conveyor belts flow shop from static and dynamic rescheduling. After analyzing the defects in coding and iteration in some algorithms in soluting scheduling problems, an artificial fish swarm algorithm based on polar coordinate coding is proposed to solute the models. Comparing the solution results shows the effectivness of models.The important research results are as follows: Through researching the conveyor belt flow shop and the LP, scheduling models of conveyor belt flow shop using JIT (CBFS-JIT) in ideal information environment is proposed. In the ideal information environment, through researching the physical structure and production mode of conveyor belt flow shop, technology of JIT and“Zero Wait”in LP and current scheduling models of flow shop, point out that these models couldn’t describe the line accurately. Aiming at the characteristics of unified speeds of the WIPs on an assembly line, preparing times needed for changing models, continuously operation of same class of WIPs in one assembly line and so on, CBFS-JIT is proposed. Then based on the parallel lines, identical parallel conveyor belt flow shop with ideal information is proposed.After making the conveyor belt flow shop as the research object and analyzing the current identification and gathering technologies, a frame and a technology of identification base on integration of RFID and barcode is designed. According to the importance of production information transparency in implementing LP in a special assembly line, an identification information acquisition frame of integrating RFID and barcode technology (IFRB) is designed. In this frame, parts and WIPs (Work in Processes) is managed in granularities, due to their species and information requirements. Based on this frame, an automatic identification technology based on integration of RFID and barcode (ITRB) is proposed, which is study in two levels: physical integration and data integration. The identification technology could support the production information transparency in manufacturing.Considering the unstable information production information environment, a kind of scheduling model of conveyor belt flow shop with unstable information is proposed. Then based on the parallel lines, an identical parallel scheduling model is proposed. Aiming at the time warp and leakage in information gathering in the application of ITRB on conveyor belt flow shop, combining the limitations in motors of conveyor belt, a kind of scheduling model using JIT with non-ideal circumstances (CFJ-NC) is proposed. Then based on the parallel lines, an identical parallel scheduling model is proposed (PCFJ-NC).To the eliminate defects of some algorithms in solving scheduling problems, a improved artificial fish swarm algorithm based on polar coordinate coding (PC-AFSA) is proposed. Aming at the defects in coding and iteration in some algorithms in solving scheduling problems and inspired from the spirit of probability coding, polar coordinate coding is proposed. After introducing polar coordinate coding into artificial fish swarm algorithm (AFSA), the three behaviors in AFSA, namely prey behavior, swarm behavior and follow behavior, are redefined in PC-AFSA. Gray decoding is proposed according to the solution process in scheduling problem. Then PC-AFSA is used to solute some optimization problems to verify its effectiveness and reliability.Solute the proposed scheduling models in PC-AFSA. Introduce the details of key technologies in solution process, including encoding and coding, iteration of matrix, distance between artificial fish, strategy of presering and the strategy of violation value. Compare the solution effects between different parameters, and choose the best parameters to solute the the scheduling models and do contrast experiments with AFSA. The simulation results show that an obvious deviation would be found of using an ideal scheduling model to describe a real production process. The model established in unstable information environment could describe the scheduling with LP in conveyor belt flow shop more accurately.Analyse the emergencies in excutiving scheduling scheme, study the rescheduling in conveyor belts flow shop in static and dynamic rescheduling. Research the impacts of machine failures and rush order in identical parallel conveyor belt flow shop. Then compare the solution results between the rescheduling and pre-scheduling, and indicate the significance of rescheduling in conveyor belt flow shop.The researches support the scheduling problems based on LP in the assembly line with conveyor belt though theories and applications. Improve the research content, technology and methods in flow shop. Provide an effective way in improving scheduling levels and the production efficiency and cutting down manufacturing costs.更多还原