基于小波包熵和ISODATA的滚动轴承故障诊断

摘要

针对滚动轴承不同振动信号各个频带的能量不同的特点,提出一种基于小波包熵( WPE)和ISODATA的集合型故障诊断方法(WPE - ISODATA).文中对滚动轴承振动信号进行采样；利用小波包提取滚动轴承振动信号的能量特征并归一化运算,将归一化的能量特征作为振动信号的概率分布进行信息熵运算,提取滚动轴承振动信号特征；以信息熵作为ISODATA聚类算法的输入进行故障辨识.滚动轴承实验结果表明:基于小波包熵和ISODATA的集合型故障诊断方法鲁棒性好,可靠性高.%An ensemble approach ,called WPE-ISODATA, based on wavelet packet entropy ( WPE) and ISODATA is presented according to every frequency band energy of different vibration signal being different in rolling bearing. The vibration signal of rolling bearing is sampled in the paper. Wavelet packet is used to extracte the energy characteristics of the vibration signal, and the extracted signal is normalized. The normalization energy characteristics is used as the probability distribution to do the information entropy calculations, to extract the vibration signal characteristics of rolling bearing. The information entropy is as the input of ISODATA clustering algorithm to fault detection. The actual bearing data were provided in the experiment, and the results showed that the ensemble approach based on wavelet packet entropy and ISODATA has good robustness and high reliability.