基于小波分析和信息融合技术的故障诊断  

A Method for Fault Diagnosis in Analog Circuits Based on Neural Network Information Fusion Technique and Wavelet Analysis

作　　者：刘美华[1] 彭良玉[1] 

机构地区：[1]湖南师范大学物理与信息科学学院,湖南长沙410081

出　　处：《微电子学与计算机》2009年第1期166-168,共3页Microelectronics & Computer

基　　金：湖南省自然科学基金项目(06JJ50117)

摘　　要：提出了利用两类不同电量信息——动态电源电流和输出电压经神经网络信息融合的模拟电路故障诊断新方法.该方法先对两类不同电量信号采样后进行小波分解,提取其在不同频段的能量,经归一化和向量关联后作为特征向量输入神经网络进行训练诊断.实验结果表明该方法比利用单一信号进行诊断效果更佳,具有快速定位、分辨率高等特点.A new method for fault diagnosis of analogue circuits based on neural network information fusion technique about dynamic supply current and output voltage is presented. Firstly, energy fault features of two kinds of different sampied signal are extracted, then they are normalization preprocessed and vector associated, finally, they are inputed neural network and trained. Simulation results illustrate that this method is better than single signal fault diagnosis and has many powerful features, such as diagnosing and locating faults quickly and exactly.

关 键 词：小波分析 神经网络 信息融合 模拟电路 故障诊断 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]