【作者】 王共冬；

【导师】 刘文剑；

【作者基本信息】 哈尔滨工业大学 ， 航空宇航制造工程， 2007， 博士

【摘要】 计算机辅助工艺设计(CAPP)是当今CAD/CAM领域的研究热点之一,可有效缩短产品工艺设计和生产的周期,对产品快速开发过程的实现具有重要意义,但对于先进复合材料领域还没有实用的系统出现。针对树脂基复合材料RTM成型工艺与热压罐成型工艺等六种成型工艺的特点,本文在分析国内外有关复合材料计算机辅助成型工艺研究成果的基础上,对树脂基复合材料成型工艺自动化、智能化设计进行了深入的探索,充分利用计算机最新发展技术来提高复合材料成型工艺的自动化水平,从根本上提高复合材料制件的批生产能力、降低生产成本。本文主要研究工作如下:在分析树脂基复合材料成型工艺过程的基础上,提出了树脂基复合材料计算机辅助成型工艺设计系统(CM-CAPPS)的总体方案,建立了树脂基复合材料计算机辅助成型工艺设计信息模型,并研究了系统的可定制性。提出了基于编码系统的树脂基复合材料制件聚类方法。在分析树脂基复合材料制件特征的基础上,建立了18位分类编码系统。在此基础上,采用粗糙集属性约简理论对制件聚类属性进行约简,建立最近邻聚类模型,并采用AIC准则对聚类结果进行评估。提高了制件聚类的速度且解决了制件聚类模型的类中心和类别数不易确定的问题。研究了树脂基复合材料成型工艺决策方法。在分析影响树脂基复合材料成型工艺因素的基础上,建立了成型工艺的评估模型;采用粗糙集属性约简理论对成型工艺评判指标进行约简,并采用等价闭包法和信息熵理论来确定指标的权值。实例证明评判结果更合理和客观。研究了基于启发式知识的树脂基复合材料层合板铺层参数优化算法。在复合材料强度分析的基础上,以蔡-希尔(Tsai-Hill)强度准则建立自适用遗传算法的适应度函数,采用带置信度的IF-THEN模糊表达方式表达层合板铺层启发式知识,在自适用遗传算法中定义了铺层规则算子,种群在每次迭代过程中进行铺层规则算子校核。算例表明该算法收敛速度较快、实用性好。建立了树脂基复合材料成型工艺知识库。研究了树脂基复合材料成型工艺中的决策知识和事例知识的构成和表达形式。提出了基于粗糙集和决策树理论的复合材料成型工艺决策知识获取算法,采用知识相依性对决策数据进行约简同时建立决策树,可以实现成型工艺决策知识的自动获取。研究了事例推理机制,对事例检索算法进行改进,采用粗糙集属性重要性对事例索引属性进行约简,并确定索引属性的权值,提高了事例检索的速度和正确性。基于上述关键技术,利用Delphi、VC面向对象编程语言和SQL Server数据库系统基础上完成原型系统开发,初步验证了上述关键技术的可行性。更多还原

【Abstract】 Computer aided process planning (CAPP) is the key techniques of CAD/CAM, which may shorten the period between product design and product manufacture effectively, and is import to rapid product design, but CAPP system has not applied to the advance resin composite material. According to the features of six methods of molding process, such as RTM (Resin Transfer Moulding) process, hot compressing molding process and so on, the automatic process and intelligent design of composite materials were studied thoroughly in this paper, based on the home and aboard research work of the computer aided composite material process planning techniques.The automatic level of manual-operated composite molding technology was completely improved by adopting the latest achievements of computer information processing technique and hence batch productivity is increased and batch cost is reduced. The following research works were included:The overall plan of CM-CAPPS (Computer Aided Composite Material Process Planning System) is proposed by analyzing of resin composite process. The information models of CM-CAPPS are established and the custom-built technique is studied.The resin composite material part clustering method based on the part code system is presented. According to the analysis of resin composite part features,a classification and coding system, which has eighteen codes, is set up. Firstly, reduction theory of rough set is utilized to reduce the features of composite part. Secondly, the nearest neighborhood clustering model of composite part is proposed. Finally, the AIC criterion is used to evaluate the clustering results. The method improves the clustering speed and solves the problem that the sort number is difficult to make certain when build the class center of clustering model.The decision method of resin composite process is studied. The comprehensive evaluation model is established through analyzing the factors of resin composite process. Reduction theory of rough set is utilized to reduce the evaluate features and the weight is determined by using equivalence closure and entropy. Finally, the case study shows that the decision method is feasible and efficient.The composite laminate stacking sequence parameter optimization based on design heuristics is studied. Through analyzing of composite laminate strength, a fitness function of adaptive genetic algorithm based design heuristics is established by using Tsai-Hill rule. Design heuristics are expressed by fuzzy IF-THEN with believe ratio, design heuristics arithmetic operators are defined in adaptive genetic algorithm and the arithmetic operators is performed in each iteration. An example is given to verify that the convergence speed is faster and the optimization result is valid.The resin composite process knowledge database is established. The structure and expression of decision and case knowledge are discussed. Based on the rough set and decision tree algorithm, the composite decision mining algorithm is presented by using knowledge dependence.The case reasoning mechanism is studied and the case retrieving algorithm is improved by using feature importance of rough set to reduce the case index features and determine the weights of case index, thus the case retrieving speed and preciseness is improved.Base on the above-mentioned key technologies; prototype system was developed via Borland Delphi, VC and Microsoft SQL server, which validated the feasibility of the technologies.更多还原