【作者】 刘长安；

【导师】 李剑峰；

【作者基本信息】 山东大学 ， 机械制造及自动化， 2003， 博士

【摘要】 随着制造技术的发展,CAPP技术受到广泛的重视。本文分析了当前CAPP研究中存在的问题,研究了CAPP中的关键技术问题,并结合订单式离散型制造企业信息化问题研究了CAPP在一体化系统中的实现方法,开发了PDM/CAPP/MRPⅡ一体化系统。论文的主要研究工作如下: 为解决CAD与CAPP之间的信息集成问题,研究了基于几何CAD数据模型的二维工程图加工特征识别方法,分析了工程图中的语义信息,基于智能模式识别理论,以语义和基本实体为模式基元,以实体之间的关系为操作规则,基于文法结构建立模式类,通过对基元的提取与句法分析与匹配实现特征识别。 论文研究了基于实例归纳的工艺设计知识自动获取方法,提出基于动态分类思想的工艺规划网络的自学习问题,分析了基于相似性原理的成组技术的事先分类存在的缺陷,基于人工智能的机器学习的原理,研究了工艺过程状态空间描述、零件状态描述、差别特征和动态聚类标准,通过对工艺设计过程的自学习实现工艺规则和工艺规划网络的学习和自动分类。学习过程中自动根据聚类标准形成若干类以及关于每个类的特征描述。工艺设计过程中,系统在获得待加工零件特征的基础上,自动匹配类型特征启用相应的工艺规划网络和规则进行工艺规划。 研究了适合于工艺知识自学习的工艺决策模型和决策策略,提出基于三级工艺规划网络的工艺决策模型,提出基于分级规划策略完成工艺路线以及工艺规程的设计。 论文研究了CAPP/MRP的并行决策模式,把交货期与动态资源负荷作为约束,基于约束推理算法,研究了基于单个特征的备选方案推理和满足交货期和资源负荷约束的回溯算法,保证了工艺规划的可行性。 论文结合典型订单式离散型制造企业进行了一体化系统的开发,开发了面向订单式离散型中小型制造业的PDM/CAPP/MRP/ERP一体化系统,系统采用客户/服务器C/S体系结构,使用大型数据库MSSQLSERVER2000作为统一数据平台,采用统一的流程管理。系统强调了CAPP在一体化系更多还原

【Abstract】 CAPP (computer aided process planning) has been attached importance broadly along with the development of manufacturing technology. The existing problem in current CAPP research work is analyzed. Some key technique problems in CAPP are studied. The realizing methods of CAPP in integration system are investigated for manufacture enterprise informalization problem with order and disperse features, based on which, a PDM/CAPP/MRPII integration system is developed. The main researches in the paper are:Machining feature recognition technology in 2D drawing based on CAD data model is researched in order to resolve the information-integrating problem between CAD and CAPP. The semantic information in 2D drawing is analyzed. Based on pattern recognition theory, the basic drawing entities are considered as the pattern cell and the relationship between entities is selected as operational rules. Then the pattern classes are established based on grammar structure and the universal recognition process is accomplished based on syntax analyzing and matching principles.Knowledge acquisition of process planning based on case inductive reasoning is researched in the paper. The self-learning problem of process planning network is put forward based on dynamic classification. The limitation of prior sorting of group technology based on comparability principle is discussed. Process state space description, workpiece state description, distiction features and dynamic clustering rules are studied based on matching learning principle in artificial intelligence. Learning and classification of process rules and process planning network are realized by the self- learning capability in the process planning process procedure. In the process of learning, several classes and feature descriptions of each class are generated automatically according to the clustering rules. In the process of process planning, the system will match更多还原