【作者】 白翱；

【导师】 唐任仲；

【作者基本信息】 浙江大学 ， 机械制造及其自动化， 2011， 博士

【摘要】 随着RFID、PDA、ZigBee等U-计算(也称为泛在计算或普适计算)技术的快速发展,一种新的下一代先进制造范式,——U-制造应运而生。将U-制造应用于离散生产车间中可解决车间数据断层、信息不能及时获取、异常情况或事件难以感知等典型问题,并弥补传统制造执行系统的不足(如数据采集范围有限、采集时间存在延迟等),从而实现对车间业务过程的精细化、实时、在线和远程管理。本文在国家自然科学基金(No.50675201)、浙江省重大科技专项(优先主题)工业项目(No.2008C01060-1)等资助下,着重研究了将U-制造应用于离散生产车间时所面临的几个关键问题,包括：U-制造运行环境的构建、泛在数据(典型的如泛在物流数据和泛在质量数据,本文主要关注这两类数据)的信息提取、U-制造信息的按需服务等。针对车间U-制造运行环境的构建问题,从确定数据采集节点和确定标识对象两个方面进行了分析。首先建立了企业应用U-制造的能力成熟度模型及其应用能力域影响因子采用层次分析法确定了影响因子权重,采用模糊综合评估法确定了其应用能力成熟度级别,从而确定U-制造应用范围,采用基于瓶颈的方法识别目标范围内的关键数据采集节点；接下来采用模糊德尔菲法确定了影响物料标识的关键因素,在此基础上,采用模糊C-均值聚类对同类物料进行了划分,获得需要被优先标识的物料。针对泛在物流数据的信息提取问题,将工位上固定式RFID读写器读取物料上的标签这一行为视为一次物流简单事件(可称之为RFID读事件)的触发,运用复杂事件处理技术中的事件操作符,将其映射和转换为六类物流复杂事件(包含领料出库、物料到达工位、物料报废、物料返工、物料终检、物料完工入库等事件),并给出了其聚合规则,分析了其聚合过程及原理；接下来将六类物流复杂事件和工序相结合,构建物流状态矩阵以实现单批次生产作业任务中所有物流复杂事件的存储；最后结合物流状态矩阵的两类参数(包括秩和元素),对单批次生产作业任务的物流进度进行了详细的分析和判别。针对泛在质量数据的信息提取问题,将移动式RFID读写器采集一次质量数据的结果定义为一个质量数据粒,并给出其形式化描述,分析了其中各个属性的来源。在质量合格率、质量损失率、质量返工率、质量直通率等已有指标的基础上,提出了质量稳定率、质量影响率等新指标。基于质量数据粒中的各个属性,给出了正常模式下和异常模式下各个指标的计算过程和步骤。接下来将六类质量指标和工序相结合,构建了面向单批次生产作业任务的质量状态矩阵,以全面描述和反映任务执行过程中物料的质量状态。针对制造信息的按需服务问题,首先提出了一种基于“物料-任务”的业务过程上下文动态建模技术,给出了其基本元素组成和建模的三个步骤,包括上下文模板的初始化、上下文模板在订单执行不同阶段中的变迁以及两类重要的信息载体,——物流状态矩阵和质量状态矩阵的嵌入。再次讨论了面向用户的U-制造信息服务规则描述和设定过程,在定义U-制造信息服务规则中四个组成元素的基础上,通过起草、验证和存储等三个步骤,进而以关系数据库表对信息服务规则加以保存。接下来详细讨论了典型U-制造服务信息的生成过程(或机制),以便为用户提供实时、在线、远程的生产决策支持。为指导离散制造企业在其生产车间中顺利应用上述四种方法,给出了各个方法的参考应用流程(实施过程),并以某中小汽车电机制造企业为对象,给出了四种方法的具体应用实例,从而验证了方法的可行性。最后对论文的主要研究内容进行了总结,归纳了其中的主要创新点,并对论文的后续研究方向进行了探讨。更多还原

【Abstract】 With the rapidly development and maturity of ubiquitous computing technologies (e.g. Radio Frequency IDentification-RFID, Personal Digital Assistant-PDA, and ZigBee), a novel and next generation advanced manufacturing paradigm,——Ubiquitous Computing-based Manufacturing (U-Manufacturing) is coming into being. Applying U-Manufacturing on discrete production shop floor could solve these problems such as data gap between top plan and bottom execution, unawareness of exceptional situation or event, information acquisition in a delayed manner, and cover the shortage of traditional manufacturing execution system, thus achieve lean, real-time, online and remote management of business process on shop floor. Under the supports from the National Natural Science Foudation of China (No.50675201) and Industrial Project of Major S&T Special Fund (Priority Subject) of Zhejiang province (No.2008C01060-1), several key issues to deploy or implement U-Manufacturing on discrete shop floor, including construction of U-Manufacturing operational environment, transformation of ubiquitous data (typicallly logistics data and quality data) into information and personalized information service for different users just in time were studied separately.To solve the problem of constructing U-Manufacturing operational environment, a quantitative and qualitative-combined method to determine the Data Capture Nodes (DCNs) and the Tagged Objects (TOs) was established. First, the capability maturity model of U-Manufacturing and influence factors of application capability domain were built, then Analytic Hierarchy Process (AHP) was used to determine the factors’weight whereas Fuzzy Comprehensive Evaluation (FCE) was applied to estimate the capability maturity level, thus obtain the application area of U-Manufacturing within enterprise. Furthermore, a bottleneck-based method was proposed to find out the suitable DCNs in the application area. Next, Fuzzy Delphi Method (FDM) was used to identify the key factors which decide if one material should be tagged or not. On that basis, Fuzzy C-Means (FCM) method was introduced to divide the same type of materials into two clusters, thus determine the important objects that need to be tagged.To solve the problem of transforming ubiquitous logistics data into information, a semantic analysis method based on Complex Event Process (CEP) technique was proposed. First, RFID tags were labeled on materials and readers were fixed at the workstation, the tags being read by the readers was treated as an occurrence of simple event, which could be also called RFID Reading Event (RRE). Then, five event operators of CEP were used to aggregate RRE into six Logistics Complex Events (LCEs), which were material outbound event, material workstation-arrival event, material reworking event, material scrapping event, material final-inspection event and material warehouse-entry event, and each of them was defined formally and its aggregation process was described. Next, based on the relationship between six LCEs and working processes, a Logistics Status Matrix (LSM) was constructed to store all of the LCEs for a single production lot execution. Finally, the logistics progress of single production lot was justified in detail with the parameters (including rank and element) of LSM in a real-time manner.To solve the problem of transforming ubiquitous quality data into quality information, a semantic analysis method based on quality index was proposed. First, the result of quality data acquisition using mobile RFID reader was treated as a quality data particle, its formal description was presented and its attributes’sources were analyzed. Next, six Key Quality Indexes (KQIs) including quality qualification rate, quality loss rate, quality rework rate, quality first-pass-yield, quality stability rate and quality affection rate were given out and defined, where the fisrt four indexes were existed indexes and the last two indexes were new indexes. Based on the data quality particle and its attributes, the computing process of each quality index was given out under the normal and abnormal patterns seperately. Finally, combined together with six indexes and working processes, a Quality Status Matrix (QSM) was constructed to store all of the KQIs for a single production lot execution, and the parameters (mainly element) of QSM were also analyzed to reflect the material’s quality timely.To solve the problem of providing information to different users just on demand, an information service method based on Business Process Context (BPC) on shop floor was proposed. First, a BPC dynamic modeling technique based on material and task was proposed, its basic modeling elements and three modeling steps (including initialization of BPC template, evolution of BPC template and embedding of LSM and QSM) were also presented. Next, from the user’s perspective, the elements of U-Manufacturing Information Service Rule (ISR) were given out and its description or setting processes (including drafting, verification and saving) were discussed. The relational database table was suggested to store ISR. Finally, the generation processes of typical service information were analyzed based on BPC and ISR.Basic application procedures of the above methods were given out in order to guide and assist the discrete manufacturing enterprises to use them on their production shop floor smoothly. Then, the feasibility of the proposed methods was proved with four specific cases in a small and medium-sized automobile motor manufacturer.In the end, all of the above researches were summarized and there main innovative points were given out, several possible research directions of U-Manufacturing which should be pay special attention to were also discussed.更多还原