产品追踪过程中失效UID恢复与补救技术研究

发布时间：2018-02-22 21:18

本文关键词： 产品追踪恢复与补救标识特征设计多信息融合预测与补救　出处：《西北工业大学》2016年博士论文　论文类型：学位论文

【摘要】：为满足不断提高的复杂产品安全管理和质量保障的要求,产品追踪是实现准确的产品技术状态管理以及快速质量追溯有效的解决途径,实现产品追踪的最直接有效手段是对产品生命周期的各阶段进行零部件的唯一标识(Unique Identification,UID),国际多家研究所与企业联合成立DPM(Direct Part Marking,DPM)工作组,选定数据矩阵码(Data Matrix,DM)作为零部件的唯一标识。然而,由于离散制造企业生产环境和工艺的复杂性容易引起DM码的磨损、污染和腐蚀等,这时产品标识难以依靠现有的识读算法自动识别,即标识失效,导致追踪断路。因此,研究产品标识失效的恢复和补救技术就变得非常必要,目前国内外大部分关于产品失效标识恢复补救的研究主要集中在工艺、图形图像等单个角度进行研究,尚未有一套产品追踪过程的失效标识恢复补救理论体系和技术框架。对于离散制造企业的产品追踪模式,依然存在包括批次产品样本信息同质化,批量生产对象标识局部或完全失效,标识失效预警和及时恢复补救等问题,现有技术手段未能较好的解决这些问题。本文以适应各种复杂应用环境的产品直接标识技术为基础,针对不同的标识失效形态和企业追踪模式存在的问题,研究产品追踪过程的标识失效恢复补救,以实现产品不间断的可靠追踪,展开了以下几个方面的研究:(1).分析了产品追踪过程标识的失效形态,给出了失效标识恢复与补救的定义,探讨了产品追踪过程中标识失效存在的问题,进一步阐述了其中的难点问题及解决途径。建立了产品失效标识恢复与补救策略,并对策略和流程模型进行了描述,从同质化产品的标识特征设计、标识失效的预测与补救和多信息融合的失效标识恢复与补救等方面展开研究,为实现产品不间断的追踪应用提供指导。(2).研究了同质化产品标识增强化特征设计及识别方法,针对批次产品样本信息之间存在同质化问题,分析了特征强化的零件直接标识,提出了基于标刻工艺方法和仿射变换的标识特征设计,并建立DM码标刻特征矩阵,设计了基于灰度共生矩阵的DM码纹理特征,并建立产品制造过程DM码纹理库。最后采用汉明距方法和欧式距离方法级联匹配目标特征和DM码标刻特征矩阵、制造过程纹理库的相似度,从而在源头增强了产品个体之间的特征差异。(3).研究了基于多信息融合的产品失效标识恢复与补救方法,由于制造企业生产环境和工艺的复杂性,导致标识发生局部或者全部失效。分析了产品制造过程的多态和时变性,依据产品物理信息、制造过程信息提出了基于多信息融合的产品失效标识恢复与补救模型,计算失效标识产品的特征数据和产品历史数据库的相似度测度,采用改进的变异系数加权法进行了初步的匹配,通过D-S证据理论分配特征的基本信度并进行融合辨识。(4).研究了基于制造历史数据的产品标识失效预测与补救方法,分析了制造历史过程大量数据与标识的关联,给出了基于DPM的制造历史信息模型,描述了标识失效的影响因素,并将制造历史数据进行主成分分析方法优化,建立了神经网络标识失效预测模型,结合神经网络预测结果和标识转移与继承方法进行了失效标识的恢复补救,避免标识失效引发的信息追踪断路。(5).将产品追踪过程失效标识恢复与补救的思想应用到实际的项目中,结合相关项目背景开发了产品制造过程数据采集与跟踪系统,对失效标识恢复与补救的方法步骤进行了企业的试运行,结合应用实例对本文所研究的理论与方法进行了验证。
[Abstract]:In order to meet the complex product safety management and constantly improve the quality assurance requirements, product tracking is to realize the product configuration management accurate and fast tracing quality effective way, the most direct and effective means to achieve product tracking is a unique identifier for each stage of the product life cycle of parts (Unique Identification, UID), International Research Institute and enterprises jointly established DPM (Direct Part Marking, DPM) working group, selected data matrix code (Data Matrix DM) as the only identification of components. However, because of the complexity of the discrete manufacturing enterprise production environment and process easy to cause the wear of DM codes, such as pollution and corrosion, then the product identification is difficult to rely on reading the existing algorithms of automatic identification, which identifies failure, resulting in tracking circuit. Therefore, research on product identification failure recovery and remediation technology becomes very necessary at present Most of the domestic and foreign research identifies product failure recovery recovery mainly in technology, graphic images of single point of view, there is not a series of product tracking and recovery recovery theory system and technical framework of the identification process. Failure for discrete manufacturing enterprise product tracking model still exist including the homogeneity of the sample batches of products information, batch production the object identifier identifies the partial or complete failure, failure warning and remedial problems in a timely manner, the existing technical means to solve these problems. In order to adapt to a variety of complex application environment, the product direct identification technology as the foundation, according to the different failure modes and business identification tracking mode existing problems, research on product tracking and identification of failure process in order to achieve remedial, reliable tracking products uninterrupted, carried out as follows: (1) analysis of the production. The failure form of goods tracing identification process, gives the definition and identification of failure recovery recovery, and explore the product tracking process identifies failure problems, further elaborated the main problems and solutions. A product identification failure recovery and recovery strategy, and the strategy and process model are described, from the identification feature the design of homogeneous products, research on prediction and identification of failure recovery and multi information fusion identification and failure recovery recovery etc., to provide guidance for the application of uninterrupted tracking products. (2). Study the homogenization of the product logo design and enhanced feature recognition method for homogenization problems between batches of products the sample information, analyzes the characteristics of strengthening parts direct identification, proposed identification feature design method and process of marking based on affine transformation, and the establishment of DM code marking feature The design of DM code matrix, texture feature based on gray level co-occurrence matrix, and the establishment of product manufacturing process DM code texture. Finally using Hamming distance method and Euclidean distance matching method of cascade target feature and DM code marking feature matrix, texture similarity of the manufacturing process, so as to enhance the differences between individual products in the source. (3). The multi information fusion of product failure recovery and recovery method based on identification, due to the complexity of manufacturing enterprise production environment and process, leading to identification of local or all fail. Analysis of the product manufacturing process and time variant polymorphism, according to physical information, manufacturing process information in the proposed multi information fusion products the failure recovery and recovery identification model based on similarity measure to calculate failure identification product data and product features of the historical database, using the improved variation coefficient weighted method A preliminary match, the basic reliability distribution characteristics and D-S evidence theory fusion identification. (4). Study on prediction and remedy for failure of manufacturing product identification based on historical data, analyses the relationship between the historical process of manufacturing large amounts of data and identification, making history information model is given based on DPM, describes the influence factors of identification failure, and will make the historical data of the principal component analysis method of optimization, to establish a prediction model of failure of neural network identification, prediction and identification of transfer and inheritance method of remedial failure identification combined with neural network, to avoid the failure caused by the identification information tracking circuit. (5) the product tracking process. The failure recovery logo to the actual project and the idea of remedial application, combined with the project background of the development of the product manufacturing process data acquisition and tracking system for failure identification. The method of recovery and recovery is carried out in the trial operation of the enterprise, and the theory and method studied in this paper are verified with the application examples.

【学位授予单位】：西北工业大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TB497

【相似文献】