变压器振动法故障诊断技术的研究

发布时间：2019-07-03 20:51

【摘要】：电力变压器是一种静止的电气设备,用来将某一数值的交流电压变成频率相同的另一种或几种数值不同的电压的设备。它是电力系统中的关键设备,其是否平安可靠运行关系着整个电网的平安运行。近些年来,基于振动信号来对变压器进行故障诊断的方法已经初具雏形,本文是通过分析传感器传输过来的振动信号来进一步的提取特征量,并依据提取到的特征量来评估变压器的工作状况。大量实验结果显示:当变压器发生故障的时候,其振动信号并不是处在平稳状态,而是非平稳状态。如何对非平稳信号进行分析,本文采用小波分析方法来进行分析,但是目前基于小波分析的变压器振动信号故障诊断方法存在一个缺陷,就是它有一个前提条件,认为小波分解的系数符合高斯分布,在实际情况中并非如此,小波变换存在着一种压缩特性,大多数的能量只是存在于一小部分的小波系数中,这与高斯分布的特性有明显的差异。本文首先对变压器故障的产生机理进行阐述,提出小波变换的分析方法,总结了小波变换的理论基础知识,然后,建立变压器振动模型,将信号分成适当的几种故障类型。在进行故障诊断时,先采用能量谱故障分析方法对信号进行故障诊断。然后建立小波域的广义高斯模型,根据广义高斯分布的概率密度函数对信号进行拟合,提取小波系数的广义高斯分布的参数作为特征量对信号进行故障诊断并结合支持向量机对信号进行了分类。之后在考虑小波系数相关性的时候,结合能量谱和建立的小波域隐性马尔可夫模型对信号进行故障诊断。最后在VC上实现小波变换和相应能量的计算,并且在VC上实现故障信号的诊断。
[Abstract]:Power transformer is a kind of static electrical equipment, which is used to change a certain value of AC voltage into another or several kinds of voltage with the same frequency. It is the key equipment in the power system, and its safe and reliable operation is related to the safe operation of the whole power grid. In recent years, the method of fault diagnosis of transformer based on vibration signal has begun to take shape. In this paper, the feature quantity is further extracted by analyzing the vibration signal transmitted by sensor, and the working condition of transformer is evaluated according to the extracted feature quantity. A large number of experimental results show that when the transformer fails, the vibration signal is not in a stationary state, but in a non-stationary state. In this paper, wavelet analysis method is used to analyze the non-stationary signal, but there is a defect in the fault diagnosis method of transformer vibration signal based on wavelet analysis, that is, it has a prerequisite. It is considered that the coefficient of wavelet decomposition accords with Gaussian distribution, which is not the case in practice, and there is a compression characteristic of wavelet transform. Most of the energy only exists in a small part of wavelet coefficients, which is obviously different from the characteristics of Gaussian distribution. In this paper, the mechanism of transformer fault is described, the analysis method of wavelet transform is put forward, and the theoretical basic knowledge of wavelet transform is summarized. Then, the vibration model of transformer is established and the signal is divided into several appropriate fault types. In the fault diagnosis, the energy spectrum fault analysis method is used to diagnose the signal. Then the generalized Gaussian model in wavelet domain is established, and the signal is fitted according to the probability density function of the generalized Gaussian distribution. The parameters of the generalized Gaussian distribution of the wavelet coefficient are extracted as the characteristic quantity to carry on the fault diagnosis to the signal, and the signal is classified by the support vector machine. Then, when considering the correlation of wavelet coefficients, the energy spectrum and the hidden Markov model in wavelet domain are combined to diagnose the fault of the signal. Finally, wavelet transform and corresponding energy calculation are realized on VC, and fault signal diagnosis is realized on VC.
【学位授予单位】：哈尔滨工程大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM41

【参考文献】