旋转机械故障特征提取新技术研究与应用

发布时间：2018-07-12 16:27

本文选题：旋转机械 + 振动分析　；参考：《华北电力大学（北京）》2011年硕士论文

【摘要】：随着现代化工业及科学技术的迅猛发展,旋转机械在工业领域也呈现出巨大的变化,并起着越来越重要的作用。尤其是电力工业中的主要机械设备和辅机正向着大型化、自动化、高效率、机电一体化等方向发展,影响安全的因素也逐渐增多。因此,要保证这些大型旋转机械安全,经济运行,旋转机械故障特征提取技术成为研究重点。本文主要研究了自回归模型(Autoregression Model,简称AR模型),小波分析,短时傅里叶变换(Short-Time Fourier Transform,简称STFT)、维格纳-威尔分布(Wigner-Ville Distribution,简称WVD)和希尔伯特黄变换(Hilbert-Huang Transform,简称HHT),并开发了旋转机械振动信号分析系统。针于AR模型,主要研究了如何确定模型的阶数,以及自相关估计、Burg法和改进的协方差法的分辨率对比,并采用轴承局部故障信号和齿轮故障信号,讨论AR模型参数估计功率谱,结果发现能得到分辨率和方差性能较好的光滑谱线,能有效的提取故障特征。本文研究了小波变换的基础理论；研究了它在旋转机械的奇异性信号,多种混合信号和含噪信号中的应用；并采用轴承局部故障信号和齿轮故障信号,讨论小波分析在特征提取中的应用,最后发现小波分析可以很好地应用在旋转机械故障信号特征提取中。为了能提取信号频率随时间的变化信息,研究了时频分析技术中的STFT、WVD和HHT的理论,讨论了STFT和WVD与傅里叶变换的区别,并研究了STFT和WVD各自的特点：STFT的分辨效果受窗函数的影响,WVD分析多分量信号时受交叉项的的干扰。研究了HHT中Hilbert变换引起的端点效应,并采用周期延拓和对称延拓两种方法抑制端点效应。本文对三种时频分析技术进行了对比,并将其应用在旋转机械振动信号的特征提取中,验证了时频分析技术可以得到信号的频率随时间变化的信息。最后,采用C++Builder和Matlab相结合的方法,开发了一个旋转机械振动信号分析系统,可以对信号进行自相关估计,Burg法估计,改进的协方差法估计,小波分析,STFT, WVD和HHT。其中,STFT, WVD和HHT是通过采用C++Builder调用Matlab引擎库中的短时傅里叶变换函数,维格纳-威尔分布函数,希尔伯特黄变换函数来实现的。
[Abstract]:With the rapid development of modern industry and science and technology, rotating machinery has shown great changes in the field of industry and plays an increasingly important role. Especially in the power industry, the main mechanical equipment and auxiliary machines are developing towards the direction of large-scale, automation, high efficiency, electromechanical integration and so on, and the factors affecting safety are also increasing gradually. Therefore, to ensure the safety and economic operation of these large rotating machinery, the fault feature extraction technology of rotating machinery has become the focus of research. In this paper, we mainly study the Autoregression Model (AR Model), wavelet analysis, Short-time Fourier transform (STFT), Wigner-Ville Distribution (WVD) and Hilbert-Huang transform (HHT) have been developed. Based on the AR model, this paper mainly studies how to determine the order of the model and the resolution comparison between the autocorrelation estimation Burg method and the improved covariance method, and discusses the power spectrum estimation of the AR model parameters by using the bearing local fault signal and the gear fault signal. The results show that smooth spectral lines with better resolution and variance can be obtained and fault features can be extracted effectively. In this paper, the basic theory of wavelet transform is studied, the application of wavelet transform in singularity signal, mixed signal and noisy signal of rotating machine is studied, and the bearing local fault signal and gear fault signal are adopted. The application of wavelet analysis in feature extraction is discussed. Finally, it is found that wavelet analysis can be applied to feature extraction of fault signals of rotating machinery. In order to extract the information of signal frequency varying with time, the theory of STFTWVD and HHT in time-frequency analysis is studied, and the difference between STFT and WVD and Fourier transform is discussed. The characteristics of STFT and WVD are studied respectively. The resolution effect of WVD is influenced by the window function and the crossover in the analysis of multi-component signals by WVD is studied. The endpoint effect caused by Hilbert transform in HHT is studied, and two methods, periodic continuation and symmetric continuation, are used to suppress the endpoint effect. In this paper, three kinds of time-frequency analysis techniques are compared and applied to feature extraction of vibration signals of rotating machinery. It is verified that time-frequency analysis technology can obtain information of signal frequency varying with time. Finally, a vibration signal analysis system for rotating machinery is developed by combining C Builder and Matlab. The system can be used to estimate signals by autocorrelation estimation, improved covariance method, wavelet analysis, STFT, WVD and HHT. Among them, STFT, WVD and HHT are realized by using C Builder to call short time Fourier transform function, Wigner distribution function and Hilbert yellow transform function in Matlab engine library.
【学位授予单位】：华北电力大学（北京）
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH165.3

【参考文献】