经验小波变换理论及其在机械故障诊断中的应用研究

项目名称： 经验小波变换理论及其在机械故障诊断中的应用研究

项目编号： No.51505002

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 机械、仪表工业

项目作者： 郑近德

作者单位： 安徽工业大学

项目金额： 20万元

中文摘要： 将故障模式从振动信号中分离并估计各个模式的瞬时特征是旋转机械故障诊断的关键。小波变换具有快速算法和分辨率高等优点，在机械故障诊断中得到了良好的应用。但小波变换对故障模式的分离不具有自适应性。经验小波变换（Empirical wavelet transform，简称EWT）以Fourier谱分割为基础，建立一种完全自适应的小波变换，自适应地将一个复杂的多分量信号分解为若干个具有紧支集Fourier谱和瞬时频率具有物理意义的调幅调频信号之和，非常适合处理机械故障振动信号。本项目拟在EWT理论的基础上提出经验谐波小波变换，广义经验小波变换和基于同步压缩变换的时变模态分解等方法，并将EWT理论引入到机械故障诊断领域，建立系统的基于EWT理论的机械早期故障和变工况机械故障诊断新方法。项目研究成果对于小波分析理论和机械故障诊断、特别是变工况机械故障诊断水平的提高都具有重要意义。

中文关键词： 机械故障诊断；经验小波变换；时频分析；变工况

英文摘要： Separating fault modes from vibration signal and estimating their instantaneous features are the kernels of rotating machinery fault diagnosis. Wavelet transform has high resolution and fast algorithm and has a good application in the field of mechanical fault diagnosis. However, separation of failure modes by using wavelet transform is not adaptive. Empirical wavelet transform (EWT) is based on the Fourier spectrum segmentation and is to build a full adaptive wavelet transform. By using EWT, a complex multi-component signal can be adaptively decomposed into several amplitude-modulation and frequency-modulation (AM-FM) signals, which have compact support Fourier spectrum and physically meaningful instantaneous frequencies. Therefore, EWT is very suitable for analyzing mechanical fault vibration signal. Based on the EWT theories, this item draw up to propose several new methods including empirical harmonic wavelet transform, generalized EWT and synchrosqueezing transform based time-varying mode decomposition and then the proposed EWT theories are introduced to mechanical fault diagnosis to build a systematic EWT theories based machinery fault diagnosis methods for incipient and variable working condition fault diagnosis. The research results of new project are of great significance for the improvement of mechanical fault diagnosis, especially for fault diagnosis under variable working conditions.

英文关键词： mechanical fault diagnosis;empirical wavelet transform;time-frequency analysis;variable working condition