非平稳旋转机械故障诊断中的时—频阶比分析研究

发布时间：2018-03-06 02:22

本文选题：旋转机械　切入点：非平稳信号　出处：《北京交通大学》2012年硕士论文　论文类型：学位论文

【摘要】：旋转机械故障诊断的一个重要内容是对间歇性非平稳故障的检测与分析。阶比分析是一种针对旋转机械非平稳振动信号的变频采样分析技术,分为硬件阶比分析与时-频阶比分析。时-频阶比分析无需特定复杂的硬件设备,克服了传统阶比分析价格昂贵和安装不便的问题。但是硬件的缺失与软件算法对时一频阶比分析产生了一定限制,使得其分析精度与速度目前与传统硬件阶比分析仍然存在较大的差距。本文在对旋转机械典型故障特征、振动信号的测量、非平稳信号的时频分析和传统阶比分析方法进行详细介绍的基础上,对新兴的非平稳旋转机械故障诊断方法时-频阶比分析的实现过程及其软件算法精度的提高进行研究。针对其中的关键问题如瞬时频率估计、峰值搜索、转速曲线拟合及奇异点平滑处理等进行改进。另外,将改进后的时-频阶比分析与Gabor展开算法相结合,实现了复杂机械振动信号中特定阶比分量的提取及其时域波的形近似重构。改进瞬时频率估计方法,将现代时频分析方法Wigner-Ville分布与短时傅里叶变换进行对比分析,研究利用Wigner-Ville提高时-频阶比分析的瞬时频率估计精度。针对多分量振动信号Wigner-Ville分析过程中产生的交叉干扰项,采用同时在时间轴和频率轴加窗的方式进行平滑处理,减小干扰项“伪峰值”对瞬时频率估计的干扰。采用一种改进的三次样条多项式拟合法对转速曲线进行拟合,同时采用一阶差分法对拟合过程中出现的奇异点进行平滑处理,提高转速曲线拟合精度,改善非平稳信号的频谱分析中存在的“频谱模糊”现象,提高了阶比谱的分辨率。利用Gabor变换算法的可逆性,将之与改进的时-频阶比分析算法相结合,在时频域内提取复杂机械振动信号中的特定阶比分量,并对其时域波形进行近似重构,实现了基于正交离散Gabor展开的时变带通阶比滤波,克服了传统阶比分析无法在时域内提取阶比分量的不足。最后,利用ST-5000A型多功能柔性转子试验台试验系统进行旋转实验,验证本文方法的正确性和有效性。
[Abstract]:The detection and analysis of intermittent non-stationary faults is an important part of rotating machinery fault diagnosis. Order ratio analysis is a frequency conversion sampling analysis technique for non-stationary vibration signals of rotating machinery. It is divided into hardware order analysis and time-frequency order analysis. It overcomes the problems of expensive and inconvenient installation of traditional order analysis, but the lack of hardware and software algorithm have some limitations on time-frequency ratio analysis. So there is still a big gap between the precision and speed of the analysis and the traditional hardware order analysis. In this paper, the typical fault characteristics of rotating machinery, the measurement of vibration signals, the time-frequency analysis of non-stationary signals and the traditional methods of order analysis are introduced in detail. In this paper, the implementation process of time-frequency ratio analysis and the improvement of the accuracy of software algorithm for fault diagnosis of non-stationary rotating machinery are studied. The key problems such as instantaneous frequency estimation, peak value searching, and so on, are discussed. The speed curve fitting and singular point smoothing are improved. In addition, the improved time-frequency order analysis is combined with the Gabor expansion algorithm. The extraction of specific order component from complex mechanical vibration signal and the shape approximate reconstruction of time-domain wave are realized. The method of instantaneous frequency estimation is improved, and the Wigner-Ville distribution of modern time-frequency analysis method is compared with that of short-time Fourier transform. In order to improve the instantaneous frequency estimation accuracy of time-frequency ratio analysis, Wigner-Ville is used to improve the instantaneous frequency estimation accuracy. In view of the cross-interference in the process of Wigner-Ville analysis of multi-component vibration signals, a smoothing method is adopted, in which both the time axis and the frequency axis are windowed at the same time. Reduce the interference of the interference term "pseudo peak" to the instantaneous frequency estimation. An improved cubic spline polynomial fitting method is used to fit the rotational speed curve and the first order difference method is used to smooth the singular points in the fitting process to improve the fitting accuracy of the rotational speed curve. In order to improve the resolution of order spectrum, the phenomenon of "spectral ambiguity" in spectrum analysis of non-stationary signals is improved. By using the reversibility of the Gabor transform algorithm and the improved time-frequency order analysis algorithm, the specific order components of the complex mechanical vibration signal are extracted in the time-frequency domain, and the time-domain waveform is approximately reconstructed. The time-varying band-pass ratio filtering based on orthogonal discrete Gabor expansion is realized, which overcomes the shortage of traditional order analysis which can not extract order components in time domain. Finally, the ST-5000A multi-functional flexible rotor test system is used to carry out rotation experiments to verify the correctness and effectiveness of the proposed method.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH165.3

【参考文献】