基于PCA和PNN的模拟电路故障诊断  

Fault Diagnosis of Analog Circuit Based on PCA and PNN

作　　者：孙健[1,2] 王成华[2] 闫之烨[1] 洪峰[2] 

机构地区：[1]南京信息职业技术学院,南京210046 [2]南京航空航天大学电子信息工程学院,南京210016

出　　处：《微电子学》2014年第1期123-126,共4页Microelectronics

基　　金：国家自然科学基金资助项目(50907033)

摘　　要：为了解决模拟电路故障识别困难的问题,提出一种基于主成分分析和概率神经网络的模拟电路故障诊断方法。该方法对采集到的模拟电路故障信息进行特征提取,将提取的故障特征归一化处理后输入概率神经网络,进行训练和故障模式的分类识别。实验结果表明,该方法是有效的,具有较高的故障诊断率。In order to solve the problem with identification of fault in analog circuit, a fault diagnosis method for analog circuits was proposed based on principal component analysis （PCA） and probabilistic neural network （PNN）. In this technique, PCA was applied to extract features of the collected information about fault in analog circuit, and the extracted fault features were normalized and input into PNN for training and identification of classified fault modes. Experimental results showed that this method was effective and had higher fault diagnosis rate.

关 键 词：模拟电路 故障诊断 主成分分析 概率神经网络 

分 类 号：TN407[电子电信—微电子学与固体电子学]