基于小波变换与递归定量分析的轴承故障信号研究

发布时间：2018-05-14 02:30

本文选题：旋转机械 + 特征提取　；参考：《哈尔滨工业大学》2012年硕士论文

【摘要】：随着现代技术的发展，旋转机械正在向着高速化、自动化方向发展，对机械关键部件的运行可靠性提出了更高的要求，因而提高故障诊断的准确性显得愈发重要。故障特征提取是实现故障诊断的重要环节，能否有效提取故障信号特征直接影响故障诊断结果的正确与否，因此对旋转机械信号特征提取的研究显得十分必要。本文首先介绍了国内外对旋转机械故障信号特征提取方法研究的现状，引入相空间重构理论，，对相空间重构中两个重要参数时间延迟和最小嵌入维数的选取进行了研究，使用互信息法求时间延迟和虚假最邻近点法求嵌入维数，同时介绍了递归图和递归定量分析方法。本文以旋转机械轴承信号为研究对象，实现了基于小波变换的特征提取，基于递归定量分析的特征提取，以及基于小波变换与递归定量分析的特征提取。将提取的特征量作为输入向量，利用概率神经网络对信号故障类型进行了分类。通过旋转机械轴承故障实验，验证了递归定量分析对提取实际旋转机械振动信号故障特征的有效性。故障的分类和评估实验结果表明，小波变换与递归定量结合的方法能更有效地提取信号的故障特征。递归定量分析在振动信号故障特征提取中的有效性，为机械系统故障诊断提供了一种新的研究方式。递归定量分析过程中将故障信号的信息以简单图形的方式呈现使得故障信号直观化，在理论和实际应用上都有着重要的意义。
[Abstract]:With the development of modern technology, rotating machinery is developing towards the direction of high speed and automation, which puts forward higher requirements for the operation reliability of the key components of machinery, so it is more and more important to improve the accuracy of fault diagnosis. Fault feature extraction is an important part of fault diagnosis. Whether the fault signal features can be extracted effectively directly affects the correctness of fault diagnosis results. Therefore, it is very necessary to study the signal feature extraction of rotating machinery. This paper first introduces the present situation of research on fault signal feature extraction of rotating machinery at home and abroad, introduces the theory of phase space reconstruction, and studies the selection of two important parameters, time delay and minimum embedding dimension, in phase space reconstruction. Using mutual information method to find time delay and false nearest point method, the embedding dimension is obtained. The recursive graph and recursive quantitative analysis method are introduced at the same time. In this paper, the feature extraction based on wavelet transform, recursive quantitative analysis and wavelet transform and recursive quantitative analysis are realized. Using the extracted feature as input vector, the fault types of signals are classified using probabilistic neural networks. Through the fault test of rotating machinery bearing, the validity of recursive quantitative analysis to extract the fault characteristics of vibration signals of actual rotating machinery is verified. The experimental results of fault classification and evaluation show that the combination of wavelet transform and recursive quantification can extract the fault features more effectively. The effectiveness of recursive quantitative analysis in fault feature extraction of vibration signals provides a new research method for fault diagnosis of mechanical systems. In the process of recursive quantitative analysis, the information of fault signal is presented in a simple graphic way, which makes the fault signal intuitionistic, which is of great significance both in theory and in practice.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH133.3;TH165.3

【参考文献】