人机结合材料选择专家系统的研究

发布时间：2018-07-04 13:42

本文选题：材料选择系统 + 人工神经网络　；参考：《北京交通大学》2013年硕士论文

【摘要】：现代设计理论认为,产品的性能和成本的70%在设计阶段就被决定了。材料的选择又是机械产品设计工作中一个极其重要的组成部分。很多机械的重大失效并不是在于它的运动设计和动力设计,而在于材料的种类和牌号选择不当。近年来,越来越多的新材料被使用,以减轻产品的重量,提高产品的质量,降低产品的成本。另外,创新设计也需要材料选用方面的创新给予支持。因此如何在数量巨大的材料库中选择最优的材料就成了国内外学者研究的热点。本文研究和实现了一个人机结合智能材料选择系统,建立了材料数据库和知识库,着重研究了BP神经网络和专家系统在材料选择方面的应用。本文的工作包括以下几个方面：一、论述了传统材料选择过程中存在的问题,分析了国内外关于选材方法和专家系统的研究现状。根据选材来源不同,弹簧钢选材采用人工神经网络,螺纹连接件选材采用专家系统。二、建立知识库,系统通过对样本学习和总结归纳获取专家知识,存放在知识库中,根据知识库中的知识建立系统的选材模块,包含一个BP神经网络和一个决策树,通过提问的方式为设计人员推荐材料,并且能够对选材结果做出合理的解释。三、建立测题库,并通过测题库生成知识谱图,制定判定用户是否是专家的准则,对普通用户正常选材,对专家用户有添加选材样本的功能。四、系统采用B/S结构,所有服务都在服务器上实现,减少客户端载荷,使用户能够更加简单方便的使用。文章最后通过一个应用实例,验证了系统的有效性。本系统不仅可以给设计人员推荐合适的材料,也可以为经验不足的用户积累选材知识,具有广泛的应用价值。
[Abstract]:The modern design theory holds that the 70% of the product's performance and cost is determined at the design stage. The selection of materials is an extremely important part of the design of mechanical products. The major failures of many machinery are not its motion design and dynamic design, but the improper selection of material types and grades in recent years. More and more new materials are used to reduce the weight of the products, improve the quality of the products and reduce the cost of the products. In addition, innovative design needs the support of the innovation of material selection. Therefore, how to select the best material in a large quantity of material has become a hot spot of research at home and abroad.
In this paper, a personal computer combined with intelligent material selection system is studied and implemented. The material database and knowledge base are established. The application of BP neural network and expert system in material selection is emphatically studied. The work of this paper includes the following aspects:
First, the problems in the selection process of traditional materials are discussed, and the research status of the material selection method and expert system at home and abroad is analyzed. According to the different sources of material selection, artificial neural network is adopted in material selection of spring steel, and the expert system is adopted for the selection of threaded connection parts.
Two, the knowledge base is established, and the system is stored in the knowledge base by learning and summarizing the samples, and is stored in the knowledge base. According to the knowledge in the knowledge base, the selection module of the system is set up, including a BP neural network and a decision tree. The material is recommended for the designer by the way of questioning, and it can make a reasonable solution to the material selection results. Release.
Three, establish the test question bank, and generate knowledge spectrum through the test database, make the criterion of judging whether the user is the expert, the normal selection of the ordinary user, and the function of adding the material samples to the expert users.
Four, the system uses B/S structure, all services are implemented on the server, reducing client load, enabling users to be more simple and convenient to use.
At the end of this paper, the effectiveness of the system is verified by an application example. This system can not only recommend suitable materials for designers, but also accumulate material for users with insufficient experience, and has extensive application value.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP182;TH122

【参考文献】