基于CBR与神经网络的摩托车智能设计系统研究  

Research on Intelligent Design System of Motorcycle Based on CBR and Neural Network

作　　者：麻芳兰[1] 何玉林[1] 李尚平[2] 冯豪[1] 

机构地区：[1]重庆大学机械工程学院 [2]广西工学院,柳州545006

出　　处：《计算机工程与应用》2005年第35期204-206,共3页Computer Engineering and Applications

基　　金：国家自然科学基金项目(编号:50375161);重庆市科技攻关项目(编号:7823-10)

摘　　要：针对摩托车总体设计过程依据相似性进行局部创新设计的特点,提出了基于CBR与神经网络的摩托车智能设计系统结构。采用面向对象方法对摩托车实例知识进行表示与组织,并将神经网络引入到实例推理中,利用ART1网络实现基于特征的动态层次聚类,通过BP网络实现从聚类模板到每一实例的索引,从而提高系统的推理效率与质量,形成有效的推理系统。In general,the overall design of the motorcycle is an innovative design in local based on the similarity of the mature product.According to this,the intelligent design system based on CBR and neural network is presented.The cases of the motorcycle are expressed and organized with the object-oriented method.Meanwhile,the neural network is introduced into the CBR system.ART1 neural network is used to cluster dynamically the cases based on the features and BP neural network is applied to index each case based on the clustered template.Therefore,the reasoning efficiency of the CBR system is enhanced and an effective reasoning system is achieved.

关 键 词：摩托车 智能设计 面向对象方法 CBR 神经网络 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]