基于ICA和SVM的滚动轴承声发射故障诊断技术

AE BASED FAULT DIAGNOSIS OF ROLLING BEARINGS BY USE OF ICA AND SVM

以滚动轴承为研究对象,提出了应用独立分量分析(ICA)和支持向量机(SVM)相结合进行滚动轴承故障诊断的方法。首先,对声发射信号(AE)进行自相关预处理,突出声发射信号的非高斯成分,使AE信号较好地满足独立分量分析的前提条件。然后,应用独立分量快速算法分离故障轴承的声发射信号,提取其状态特征向量,利用支持向量机的模式识别和非线性回归功能来完成滚动轴承故障的识别。试验结果表明,利用独立分量分析方法提取的故障状态特征向量与支持向量机相结合可以有效、准确地识别滚动轴承的故障模式,为滚动轴承故障诊断提供了一种新型的方法。

With resorting to the acoustic signals acquired from rolling bearings rig, a fault diagnosis method for rolling bearings based on independent component analysis (ICA) and support vector machine (SVM) was presented. The preprocessing of autocorrelation function for ICA was proposed, by which the non-Gaussian parts of the individual components stand out in the acoustic emission (AE) signals to satisfy the ICA conditions. i.e. statistical independence of sources. The fast ICA algorithm was applied successfully to separate the acoustic emission signals of the fault rolling bearings and extract their state features. The SVM was used together to accomplish pattern reorganization and non-linear regression and to achieve the fault diagnosis of rolling bearings. The experimental results demonstrate that the fault state features extracted by ICA, combining with SVM, can recognize the fault pattern of rolling bearings effectively and correctly.