Application Study of EMD-AR and SVM in the Fault Diagnosis

Abstract:

Since the equipment early fault signal is very weak and unstable, it is very difficult to judge the fault type and fault degree through the traditional method of fault diagnosis .For this reason, this paper presents an effective method. Firstly, the fault signal is decomposed into a number of intrinsic mode functions (IMF) by Empirical Mode Decomposition (EMD) method, then the IMF’s energy entropy is calculated, the AR model is established, and the auto-regressive parameters are extracted. At the same time, the residual variance of the model is taken as the fault feature vector, which can be used as input parameters of SVM classifier to distinguish working state and fault types of the equipment. Finally, the validity of the proposed method is verified by taking the rolling bearing fault diagnosis as an example.

• Periodical: Advanced Materials Research (Volumes 631-632)
• Pages: 1357-1362
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.631-632.1357
• Online since: January 2013
• Authors: Zhen Wang, Yang Yang, Wei Xin Yang
• Keywords: AR Model, EMD Decomposition, Energy Entropy, Support Vector Machine (SVM)

References

[1] VANIA A, PENNACCHI P. Experimental and theoretical application of fault identification measures of accuracy in rotating machine diagnostics. Mechanical Systems and Signal Processing. 8 (2004) 329-352.

DOI: 10.1016/s0888-3270(03)00014-1

[2] Weiqiang Deng, Yuegang Wang, Yingtao Yang, Wenda Zheng: Improved Estimation Method for ARMA Model Parameters. Journal of Vibration Measurement & Diagnosis. 31 (2011) 378-380.

[3] Manel M R, Jose L R A. Support vector machines for nonlinear kernel ARMA system identification. IEEE Transactions on Neural Networks. 17 (2006) 1617-1622.

DOI: 10.1109/tnn.2006.879767

[4] Huang N E. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis. Proc R Soc Lond A. 454(1988), 903-995.

[5] Shuting Wan, Haixia Tong, Binghui Dong. Bearing Fault Diagnosis Using Wavelet Packet Transform and Least Square Support Vector Machines. Journal of Vibration Measurement & Diagnosis. 30(2009), 149-152.

[6] Yaguo Lei. Machinery Fault Diagnosis Based on Improved Hilbert-Huang Transform. China Mechanical Engineering. 47(2011), 71-77.

DOI: 10.3901/jme.2011.05.071

[7] Chao Zhang, Jianjun Chen, Xun Guo. A gear fault diagnosis method based on EMD energy entropy and SVM. Journal of Vibration and Shock. 29(2010), 36-40.

[8] Shuzi Yang, Ya Wu, Jianping Xuan, etc. Time Series Analysis in Engineering Application. Huazhong University of Science and Technology press, Wuhan, (2007).