基于Treelet变换的模拟电路故障诊断方法研究
发布时间:2018-04-01 00:08
本文选题:模拟电路 切入点:故障诊断 出处:《湖南师范大学》2016年硕士论文
【摘要】:随着电子电路的发展,电路集成度的不断提高,人们对模拟电路故障诊断提出了更高的要求。但由于模拟电路本身具有容差、非线性和故障现象多样性等特点,使得模拟电路故障诊断成为电路发展的热点和难点。在模拟电路故障诊断中,故障特征向量的提取和故障模式识别是研究的重点与难点。但模式识别过程中获得的原始数据往往包含大量的冗余信息,会影响故障诊断的效率和准确率。因此,模拟电路故障诊断的一个关键点是如何有效地提取模拟电路的特征向量。本文以模拟电路故障特征向量提取为出发点,研究了基于Treelet变换和混沌神经网络的模拟电路故障诊断方法。主要工作如下:(1)介绍了小波分析和层次聚类等故障特征提取方法。重点研究了一种新的模拟电路故障特征提取方法即Treelet变换。Treelet变换是一种将PCA、小波分析和层次聚类结合在一起的自适应的多尺度的数据分析方法,特别适用于高维数据的降维。通过对这几种特征提取方法进行研究和对比,证明基于Treelet变换的模拟电路故障诊断率相比其他方法要高。(2)介绍了人工神经网络的原理及应用,对BP神经网络的结构、学习算法进行了研究,针对BP神经网络收敛速度慢、容易限于局部最小的特点,提出了混沌神经网络,利用混沌的特性构造神经网络,使神经网络具有混沌特性,优化网络结构。(3)将Treelet变换与混沌神经网络结合应用于模拟电路故障诊断,故障诊断结果与BP神经网络和小波神经网络方法相比,本文方法在模拟电路故障诊断中比BP网络、小波神经网络诊断精度更高,收敛速度更快。
[Abstract]:With the development of electronic circuits and the continuous improvement of circuit integration, people put forward higher requirements for analog circuit fault diagnosis. However, the analog circuit itself has the characteristics of tolerance, nonlinearity and variety of fault phenomena, etc. In the analog circuit fault diagnosis, the analog circuit fault diagnosis has become a hot and difficult point in the development of the circuit. Fault feature vector extraction and fault pattern recognition are the key and difficult points in the research, but the original data obtained in the pattern recognition process often contain a lot of redundant information, which will affect the efficiency and accuracy of fault diagnosis. A key point of analog circuit fault diagnosis is how to extract the eigenvector of analog circuit effectively. The fault diagnosis method of analog circuit based on Treelet transform and chaotic neural network is studied. The main work is as follows: 1) the methods of fault feature extraction such as wavelet analysis and hierarchical clustering are introduced. Obstacle feature extraction method, I. E. Treelet transform. Treelet transform, is an adaptive multi-scale data analysis method, which combines Treelet, wavelet analysis and hierarchical clustering. It is especially suitable for dimensionality reduction of high-dimensional data. By studying and comparing these feature extraction methods, it is proved that the fault diagnosis rate of analog circuits based on Treelet transform is higher than that of other methods.) the principle and application of artificial neural network are introduced. In this paper, the structure and learning algorithm of BP neural network are studied. In view of the slow convergence speed of BP neural network, which is easy to be limited to the local minimum, a chaotic neural network is proposed, and the neural network is constructed by using the characteristic of chaos. Treelet transform and chaotic neural network are combined in analog circuit fault diagnosis. The fault diagnosis results are compared with BP neural network and wavelet neural network method. In the fault diagnosis of analog circuit, the wavelet neural network has higher diagnostic accuracy and faster convergence speed than BP neural network.
【学位授予单位】:湖南师范大学
【学位级别】:硕士
【学位授予年份】:2016
【分类号】:TN710
【参考文献】
相关期刊论文 前3条
1 高坤;何怡刚;于文新;谭阳红;童耀南;;基于曲波分析和ELM模拟电路故障诊断方法[J];湘潭大学自然科学学报;2014年03期
2 郭富强;;基于径向基函数网络的模拟电路故障诊断设计[J];电子设计工程;2011年09期
3 郑红岩;李红波;曾繁景;李铁峰;;基于峭度和负熵的模拟电路故障诊断方法研究[J];信息工程大学学报;2011年02期
相关博士学位论文 前6条
1 谢涛;基于(多)小波(包)、神经网络及优化的模拟电路故障诊断研究[D];湖南大学;2011年
2 袁莉芬;基于独立成分分析技术的模拟电路故障诊断新方法[D];湖南大学;2011年
3 孙永奎;基于支持向量机的模拟电路故障诊断方法研究[D];电子科技大学;2009年
4 刘美容;基于遗传算法、小波与神经网络的模拟电路故障诊断方法[D];湖南大学;2009年
5 王承;基于神经网络的模拟电路故障诊断方法研究[D];电子科技大学;2005年
6 谭阳红;基于小波和神经网络的大规模模拟电路故障诊断研究[D];湖南大学;2005年
,本文编号:1693073
本文链接:https://www.wllwen.com/kejilunwen/dianzigongchenglunwen/1693073.html