【作者】 杨宁；

【导师】 阮雪榆； 周雄辉；

【作者基本信息】 上海交通大学 ， 材料加工工程， 2005， 博士

【摘要】 可制造性评价及流程管理是计算机支持的协同工作与先进制造技术相结合的产物,是一项针对企业经营过程管理的软件和系统技术。采用工作流管理技术是企业在激烈市场竞争压力和科学技术进步推动下的必然选择,是企业高质量、短时间、低成本地推动新产品研究与开发的重要保证。注塑产品开发是一个十分复杂的过程,它涉及产品设计、模具设计、注塑成型过程计算机仿真等多个方面。以往,对注塑产品开发的研究主要集中于CAD/CAE/CAM方面,而对产品的开发工作流程管理等方面研究则较少。当涉及工作流程管理的问题时,一般是通过手动或半自动的方式进行处理,大大降低了产品开发的效率,难以适应新制造模式下企业过程管理的需求。本文将工作流管理技术引入注塑产品开发过程中,结合并行工程、敏捷制造等先进制造思想,对敏捷制造环境下注塑产品并行开发工作流管理系统关键技术进行了研究和探讨。其主要研究内容如下:根据现代产品开发的特点和需求,在对工作流管理相关技术全面回顾的基础上,提出了敏捷环境下的注塑产品并行开发工作流管理系统模型。基于对象描述方法和标准建模语言,给出了工作流元模型的形式化定义和图形化描述,使得抽象的工作流元模型更加贴近于具体应用。基于Web Service技术,建立了敏捷环境下注塑产品并行开发工作流集成框架,使得产品开发活动更容易、有效地被部署,基于工作流技术的各种应用能够实时地、无缝地进行互操作,从而保证了敏捷模式下产品开发活动的顺利进行。深入研究了并行环境下开发活动间的相互关系,首次提出了一种基于映射机制和事例推理的设计结构矩阵(DSM)计算方法。首先,解析了注塑件特征与设计结构矩阵间的关系,提出了一个包括特征映射、工艺映射和邻接映射关系的特征-活动的多阶段映射关系表达式,获得了零件特征与设计结构矩阵之间关系的定性认识。然后,以事例推理技术为量化求解平台,用模拟退火算法和粗糙集对事例的索引、检索、修改学习等关键技术进行了重点研究,并建立了数学模型及相应算法,通过多种技术的相互融合与渗透,成功继承与模拟专家的经验知识。深入研究了耦合活动对开发任务执行次序的影响,以图论的相关原理为基础,提出了一种新的基于图论的并行流程规划方法。围绕并行活动的定耦与解耦问题,将整个流程规划分为活动识别和活动规划两大阶段。在活动识别阶段,通过邻接矩阵、可达矩阵和强连通矩阵,对耦合活动进行了识别。在活动规划阶段,首先借助缩减矩阵、层次递阶矩阵,通过建立相邻层次活动间的偏序关系,对开发活动进行总体规划。在此基础上,借助模糊设计结构矩阵,通过耦合活动间关联强度大小,确定耦合活动撕裂优先级,对开发活动的进行局部规划,以最终实现并行开发流程的优化与重组。整套方法一个突出的优点和创新点是能够实现流程图的自动生成,从而大大提高了工作流程规划的自动化水平。研究了敏捷制造环境下合作伙伴的选择方法,结合模具制造企业发展的战略需求,提出了一个面向模具开发的合作伙伴模糊选择方法。首先,建立了综合评价指标体系,对敏捷开发任务需求进行了明确的描述。然后,对当前主要评价方法进行了对比与分析,面向整条生产链的优化,提出了一个面向注塑产品并行开发的合作伙伴多指标优化模型,以追求整个生产链性价比的最大化为目标。进一步以生产链作为禁忌对象,基于活动位置禁忌,提出了一个的合作伙伴优选禁忌搜索算法。针对注塑模具领域强经验弱理论的特点,各评价指标及权重值均采用三角模糊数表示,使决策能够更符合领域专家的思维习惯。研究了敏捷制造环境下生产任务调度方法,结合模具生产制造的特点,提出了一个柔性动态的产品并行开发任务调度模型,并对模型中柔性动态的环境参数进行了详细计算。在此基础上,提出了一个基于扩展工序编码方法的遗传算法,并对算法中染色体编码/解码、遗传算子设计、优选策略设计等关键技术进行了重点研究。为保证任务尽早完工,染色体编码/解码分别通过柔性和活动调度相结合的方法进行设计。为避免计算过程潜在的冲突与死锁情况的发生,染色体交叉算子通过次序交叉和“冒泡”排序相结合的方法进行设计。为避免由于染色体雷同而造成种群整体的早熟现象的出现,染色体变异算子通过差异度的概念进行设计。为保证遗传算法收敛于全局最优解,适应度函数通过调节参数进行设计,以维持种群的多样性。深入研究了工作流过程建模与监控方法,针对注塑产品并行开发流程复杂、环节多的特点,首次提出了一种带有总控页的层次赋时有色Petri网模型-CHCTPN(Control page-based Hierarchical Colored and Timed Petri Nets)。为降低开发活动及其状态管理的复杂性,在保留CTPN描述优点的基础上,CHCTPN通过层次聚类的方法来对耦合开发活动分而治之。为了有效监控复杂环境下工作流状态,保证开发任务的如期进行,CHCTPN通过将控制页引入Petri网,力求使各种过程控制功能与方法能够在Petri网中得到表达与应用。更多还原

【Abstract】 Workflow management is a research area which combines computer supported cooperative work and advanced manufacturing technology in the process of product development. Workflow management is a research direction under the pressure of market competition and forces of technology progress, provides an efficient development mode for design new product in high quality, short period and low cost.Development of injection product is a very complex process, including product design, mold design and simulation of injection process, and so on. In the past, injection product development mainly focused on CAD/CAE/CAM, and management of development workflow is poorly concerned. In general, if workflow management problems are encountered, manual or semiautomatic mode is always used but information process efficiency is barely satisfied. So it can’t match the demand of workflow management for a company. In this paper, workflow management technology is adopted in the development of injection product. With the idea of concurrent engineering and agile manufacturing, key technologies for injection products concurrent development workflow management system are deeply studied. The main content and creative points are summarized as follows:According to the requirement of modern product development, on the basis of comprehensive review of workflow management technologies, an injection products concurrent development workflow management system model is presented. By means of object-oriented method and uniform modeling language, a formalization definition and graphical description of workflow metal model are obtained, which make abstract metal model more approaching to the practice. By means of Web Service, an integrated workflow frame for injection products development is build, which make development activities and various workflow application can be more easily deployed.Relationship of concurrent development activities is deeply and systematically studied. On the basis of mapping mechanism and case-based reasoning (CBR) technology, a new algorithm for calculating the design structure matrix (DSM) is presented for the first time. On the analysis of the mapping mechanism between product feature and DSM, a synthesis mapping expression is proposed, which qualitatively demonstrates the relationship between product feature and DSM. The expression includes feature mapping, process mapping and adjacent mapping. After analyzing the technology challenge of CBR—case retrieving and adaptation, a novel CBR quantificational calculating platform is set up which is based on annealing simulation algorithm and rough set, and its mathematical model and associate algorithm are established. This platform efficiently improves the quality of case retrieving and adaptation by inheriting and simulating expert experience.The influence of coupled activities on workflow planning is deeply and systematically studied. On the basis of graph theory, a novel workflow planning method is presented. Focused on coupling and decoupling thesis, the whole process of workflow planning is divided into two phrases—activity identification and activity planning. In activity identification phrase, coupled activity can be recognized by Boolean operation of adjacent matrix, accessible matrix and strong connected matrix. In activity planning phrase, global planning result is obtained by compressing and hierarchically operating accessible matrix, and local planning result is obtained by comparing the coupled activities correlation strength of fuzzy design structure matrix. An advantage of the method is that mapping from design structure matrix to workflow graph can be automatically realized, which improves the efficiency of workflow planning greatly.Partner selection method is studied. On the basis of workflow planning, a fuzzy partner optimization selection strategy for mold development is presented. With the development strategy requirements of mold manufacturing enterprise, a comprehensive evaluating attributes set is build, which provides definite description of source data for partner evaluating. After analyzing and comparing the current evaluating methods, an optimization model for order preference is introduced for partner selection, which can ensure the maximal performance/price ratio for whole manufacturing chain. According to the character of strong experience and weak theory of mold manufacturing field, all decision parameters are expressed by extend triangular fuzzy number, which is more compatible with expert judgment habit.Task scheduling method of agile manufacturing was investigated. A flexible and dynamic scheduling model of concurrent product development was presented and dynamic environment parameter was calculated. Based on that, a genetic algorithm of extended procedure coding was presented and key technology such as coding/decoding, genetic operator, optimizing strategy was researched. To complete task as soon as possible, chromosome coding/decoding was designed with flexible and active scheduling method. To avoid potential conflicting and deadlock, chromosome crossover operator was designed with partially mapping and bubbling method. To avoid population premature because of chromosome similarity, chromosome mutation operator was designed with the concept of difference degree. To obtaining the global optimization chromosome, fitness function was designed with adjustment parameter.The methodology of workflow modeling and monitoring is deeply studied. On the basis of colored and timed Petri nets, a novel control page-based hierarchical colored and timed Petri nets-CHCTPN is presented. For reducing the complexity of activities and status managing, hierarchical planning ideal was used to divide and rule the coupled activities in CHCTPN. Furthermore, for efficiently monitoring and implementing the complex workflow, control page was introduced to express and apply practice control function in CHCTPN.更多还原