Fault diagnosis of bearing based on empirical modedecomposition and K-means clustering
基于EMD和优化K-均值聚类算法诊断滚动轴承故障

The vibration signal is nonstationary and the sample with typical fault is difficult to acquire. The severity of fault diagnosis is the same important to style diagnosis. This paper presented a novel fault diagnosis of bearings based on the characteristic fault frequency and K-means clustering. The reconstructed signal could be obtained by some set of IMF components of the vibration signal by EMD. It performed the Hilbert envelope analysis to reconstructed signal. From the power spectrum of Hilbert envelope signal, it could identify the amplitude of the characteristic fault frequency and its integer multiples, which was used to diagnose the style and severity of fault. The result demonstrates that the proposed method based on EMD and K-means clustering can recognize the style and severity of bearing fault.