基于实例的学习矢量量化神经网络诊断方法及其应用  

Hybrid CBR and learning vector quantization neural network approach and its application

作　　者：张国辉[1] 

机构地区：[1]华南理工大学汽车工程学院,广州510641

出　　处：《制造业自动化》2006年第6期11-14,共4页Manufacturing Automation

基　　金：国家863计划资助项目(2001AA423230)

摘　　要：基于CBR良好的可扩充性、可移植性和神经网络极强的分类能力,提出了基于实例的学习矢量量化神经网络诊断方法。该方法应用于机械故障诊断系统中,可以减小实例搜索空间,提高实例检索效率。论述了系统的设计方法和应用步骤。Based on good expandability and portability of case-based reasoning（CBR） and high classification capability of artificial neural network （ANN）, a hybrid CBR and learning vector quantization neural network approach is proposed. In this hybrid approach, which has been applied to a mechanical fault diagnosis system （FDS）, the learning vector quantization neural network has been incorporated to the CBR cycle to improve the efficiency and accuracy of the fault diagnosis process. The implement and application steps of this approach are detailed.

关 键 词：学习矢量量化神经网络 基于实例推理 故障诊断 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] TP206[自动化与计算机技术—控制科学与工程]