基于ITD信息熵与PNN的轴承故障诊断

Bearing Fault Diagnosis Based on ITD Information Entropy and PNN

滚动轴承是机械传动系统的重要组成部分,针对其故障率高、故障情况复杂的问题,提出一 种基于固有时间尺度分解(ITD)信息熵与概率神经网络(PNN)的故障诊断方法。首先使用ITD方法对信号 进行分解,对分解的分量进行相关系数计算,然后选取与原始信号相关系数大的前4层分量进行重构,提 取前4层分量的样本熵与能量熵,最后将提取的熵值用PNN进行故障诊断,并与支持向量机(SVM)的诊断结 果进行对比,结果表明:PNN相对于SVM可以提高故障诊断的正确率,正确率高达91.25%。

Rolling bearing is an important part of mechanical transmission system. Aiming at the problems of high failure rate and complex fault situation, a fault diagnosis method based on inherent time scale decomposition(ITD) information entropy and probabilistic neural network(PNN) was proposed. First, the signal was decomposed using the ITD method, and the correlation coefficient was calculated for the decomposed components. Then the first 4 layers with the correlation coefficient of the original signal were selected for reconstruction, and the sample entropy and energy entropy of the first 4 layers were extracted. Finally, the extracted entropy was diagnosed by PNN and compared with the diagnosis results of supported vector machine(SVM). The results show that PNN can improve the accuracy of fault diagnosis compared with SVM, and the correct rate is as high as 91.25%.