基于定量小波基选取与改进LDB算法的机械故障诊断

Rotating Machine Fault Diagnosis Based on QuantitativeWavelet Base Selection and Improved LDB

摘要:

通过研究信息论中的各种测度,提出了一种以信息论中的共信息熵来开展故障信号分析的定量小波基选取的方法;同时针对局域判别基算法(Local Discriminate Bases, LDB)自身存在的一些问题,提出了一种基于改进的局域判别基理论的最优小波基分解算法,构造了以局域判别基空间上的结点能量为元素的特征矢量,并在滚动轴承上进行了实验研究;实验分析表明,该算法可以有效地识别旋转机械系统中不同严重程度的故障,且与原始的LDB算法相比,改进后的LDB算法对提高识别率和降低计算复杂度都有着明显的优势。

Abstract:

This paper presents an effective approach for rotating machine fault diagnosis, based on quantitative wavelet selection and improved local discriminate bases. Mutual information entropy is utilized as a quantitative measure to select the most suitable base wavelet for wavelet packet transform. Then an improved local discriminate bases method based on the normalized energy difference and relative entropy is used to choose the optimal set of orthogonal wavelet subspaces. After that, two optimal sets of orthogonal subspaces from wavelet packet decomposition have been obtained and the energy features extracted from the subspaces appearing in both sets will be selected as input to the support vector machine (SVM) to diagnose the fault states of rotating machines. Experiment study conducted from a rolling bearing test setup has verified the effectiveness of the proposed method for machine defect severity evaluation. 