【作者】 孟庆智；

【导师】 王军；

【作者基本信息】 燕山大学 ， 机械制造及其自动化， 2010， 硕士

【摘要】 智能化、集成化是CAPP系统的主要发展方向,利用人工智能技术进行工艺决策是智能CAPP系统的关键技术。本文主要研究了人工智能中的计算智能技术应用于特征加工方案选择、定位基准和工艺路线决策的问题。首先,本文构建了智能化CAPP系统的体系结构,采用了基于多种人工智能技术的多知识库协同工作的方式,并应用内、外知识库的结构形式解决知识与CAPP系统、用户之间的信息联系,该体系结构也实现了CAD/CAPP的集成。在对回转体零件的加工特征及其加工方案分析的基础上,建立特征加工方案决策模型,利用BP神经网络自动决策得到零件特征的所有可行加工方案,详细研究了该神经网络的结构,输入输出处理方法、训练函数和训练样本的选取等问题。分析了回转体零件定位基准选择的影响因素,建立了基于BP神经网络的定位基准选择模型,详细讨论了该神经网络的结构,输入输出编码方法、训练函数和训练样本的选取等问题。建立了基于遗传算法的工艺路线优化模型,利用约束矩阵来表示加工元之间的约束关系,而且约束矩阵由系统自动生成,开发了用于加工元序列有效性检验与调整的算法,研究基于遗传算法的工艺路线优化过程,包括基因编码、染色体适应度函数的设计、选择、交叉和变异等。最后,利用VC++语言编制了程序,对上述3个决策模型进行了验证。对系统和部分界面及工艺决策的工作流程作了介绍,给出了工艺设计实例,验证了决策方法的正确性和可行性。更多还原

【Abstract】 Intelligence and integration are main developing directions of CAPP system. Process decision based on the artificial intelligence technology is the key technology of intelligent CAPP system. The paper mainly studys how the computational intelligence in artificial intelligence is applied in decision for machining schemes of featurs, selection for locating datum and optimization of process route.First, the architecture of intelligent CAPP system is designed, in which multiple knowledge bases based on multiple artificial intelligence technologies are co-working, and in order to achieve information linkage between knowledge and CAPP system, users, we use the inner and outer knowledge bases. The integration between CAD and CAPP is also realized.Based opon the analysis of features and machining schemes of turning parts, the decision model for machining schemes of features is put forward. By BP neural network all possible machining schemes of features can be automatically obtained. Network structure, input/output processing methods, training function and selection for training samples are traversed.Under the analysis of factors which influence selection for locating datum of turning parts, the selection model for locating datum based on BP neural network is established. Network structure, input/output processing methods, training function and selection for training samples are traversed.The optimization model for process route based genetic algorithm is built. In order to describe precedence relation among processing units, constraint matrix is introduced and it is automaticly generated by computers. The general program of constraint adjustment algorithm which make the processing units’sequence validity is developed. Optimization course of process route based genetic algorithm is dicussed, including gene incoding, design for fitness function of chromosome, selection, crossover and mutation and so on.Finally use VC++ to program three process decision models above. The system interface and decision process are shown. A turing part is used to validate them validity and feasibility.更多还原