摘 要：随着电子工业迅速发展，模拟电路的复杂度和集成度不断增长，模拟电路故障诊断的要求不断提高，其故障诊断方法的研究也更具现实意义。由于传统的模拟电路故障诊断方法和理论的欠完善，在实际工程应用中很难达到预期的效果。现代方法中，小波分析有表征信号局部特征的能力，神经网络有处理复杂模式以及联想、推测和记忆功能，二者结合为模拟电路故障诊断提供了一条有效途径，成为了研究热门。本文讲述了模拟电路故障诊断的发展、理论和方法，重点对小波分析和神经网络结合的诊断方法进行研究。并通过小波从仿真数据中提取特征向量，送入BP网络进行诊断，与BP网络单独诊断进行比较，结果显示，小波预处理对提高BP网络故障诊断效率和准确率有明显作用。75308

毕业论文关键词：模拟电路，故障诊断，神经网络，小波分析

Abstract：With the rapid development of electronic industry, bigger of the complexity and integration of analog circuits, it improved the requirements for fault diagnosis for analog circuits and more realistic meaningful that study on fault diagnosis for analog circuits。 Because of owing perfect of the traditional method and theory of fault diagnosis for analog circuits, it’s very hard to archive the expected effect in the practical engineering application。 Among modern methods, owing to wavelet analysis’ excellent performance in characterizing local signal characteristics in time and frequency domain and the ability of neural network in dealing with complex model, association, speculation and memory, the combination of neural network and wavelet analysis becomes a effective approach for analog circuits fault diagnosis。 The development, theory and methods of fault diagnosis for analog circuits are discussed in this paper and the diagnosis method of the combination of neural network and wavelet analysis is in-depth studied。 Extracting the feature vector through the wavelet。 Transfer the feature vector into the BP network 

for fault diagnosis compared with BP neural network fault diagnosis without preprocessing。 The result data show that wavelet preprocessing plays a significant role in improving the efficient and 

accuracy of BP network fault diagnosis。

Key words: analog circuits, fault diagnosis, neural network, wavelet analysis 

目   录

1 前言 3

1。1 模拟电路故障诊断的研究背景及意义 3

1。2 模拟电路故障诊断现代方法的研究成果 4

1。3 模拟电路故障诊断理论的未来发展方向及趋势 5

2 神经网络诊断理论介绍 6

2。1 模拟电路故障诊断的思想概述 6

2。2 人工神经网络理论概述 7

2。3 BP网络 9

2。4 BP神经网络故障诊断实例 11

3 小波分析理论及其应用 16

3。1连续小波变换 16

3。2 几种常用的小波 17

3。3 小波变换在故障诊断中对提取特征向量的应用 18

4 小波神经网络方法研究 22

4。1 小波与神经网络的结合方式 22

4。2 小波神经网络的故障诊断