基于BP神经网络专家系统的设计质量评价方法  

Evaluation Method of Product Design Quality Based on BP Neural Network Expert System

作　　者：刘杨[1] 任朝晖[1] 刘挺[1] 冯霏[1] 

机构地区：[1]东北大学机械工程与自动化学院,辽宁沈阳110004

出　　处：《机械与电子》2011年第2期7-10,共4页Machinery & Electronics

基　　金：国家自然科学基金资助项目(50975044)

摘　　要：以家用轿车为例,采用基于BP神经网络的专家系统对不同品牌的产品进行质量评价,确定神经网络的训练样本进行训练,以获得各神经元的连接权值和阈值,存储连接权值于一个数据文件中,形成专家系统知识库,经过推理机推理,并通过解释机制将得到的机器语言转换成用户能够理解的方式进行表达,可实现在无专家条件下对产品设计质量进行评价,进而论证了采用专家系统评价产品设计质量的可行性。With the family car as example, the expert system based on BP neural network was used to evaluated different brands of products, The neural network training samples were determined and the connection weights of neurons and thresholds were obtained. The formation of expert system knowledge was set up while the connection weights were stored in a data file. After the inference en- gine reasoning and by explaining the mechanism will he converted into the way of machine language which the user can understand the expression, to evaluate the quality of product design can be realized without the expert. It is demonstrated feasibility of the quality of product design expert system for evaluation.

关 键 词：产品设计质量 BP神经网络 神经网络专家系统 MATLAB 评价方法 

分 类 号：TH113.1[机械工程—机械设计及理论] TU66[建筑科学—建筑技术科学]