Vibration Gear Fault Diagnostics Technique Using Wavelet Support Vector Machine

Abstract:

Intelligent diagnostics tool for detecting damaged bevel gears was developed based on wavelet support vector machine (WSVM). In this technique, the existing method of SVM was modified by introducing Haar wavelet function as kernel for mapping input data into feature space. The developed method was experimentally evaluated by vibration data measured from test rig machinery fault simulator (MFS). There were four conditions of gears namely normal, worn, teeth defect and one missing-teeth which has been experimented. Statistical features were then calculated from vibration signals and they were employed as input data for training WSVM. Fault diagnostics of bevel gear was performed by executing classification task in trained WSVM. The accuracy of fault diagnostics were evaluated by testing procedure through vibration data acquired from test rig. The results show that the proposed system gives plausible performance in fault diagnostics based on experimental work.

• Periodical: Applied Mechanics and Materials (Volume 564)
• Pages: 182-188
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.564.182
• Online since: June 2014
• Authors: Achmad Widodo*, D.P. Dewi Widowati, D. Satrijo, I. Haryanto
• Keywords: Fault Diagnostics, Gearbox, Support Vector Machine (SVM), Wavelet Function

References

[1] J. Korbicz, J. Koscielny, Z. Kowalczuk, W. Cholewa, Fault diagnosis: model, artificial intelligence, applications, Springer-Verlag, New York, (2004).

[2] Y. Murphey, J. Crossman, Z. Chen, J. Cardillo: IEEE Trans. Veh. Technol. Vol. 52(4) (2003) pp.1076-1098.

DOI: 10.1109/tvt.2003.814236

[3] I. Rish, M. Brodie, S. Ma, N. Odintsova, A. Beygelzimer, G. Grabarnik, K. Hernadez: IEEE Trans. Neural Netw. Vol. 16(5) (2005) pp.1088-1109.

DOI: 10.1109/tnn.2005.853423

[4] B. Paya, I. Esat: Mech. Syst. Signal Proc. Vol. 5 (1997) pp.751-765.

[5] R. Isermann: IEEE Trans. Syst. Man Cybern. A, Syst., Human. Vol. 28(2) (1998) pp.221-235.

[6] G. Castellano, A. Fanelli, C. Mencar: IEEE Trans. Syst. Man, Cybern, B, Cybern. Vol. 34 (2004) pp.725-731.

DOI: 10.1109/tsmcb.2003.811291

[7] J. Wang, C. Lee: IEEE Trans. Fuzzy Syst. Vol. 12 (2002) pp.790-802.

[8] A. Widodo, B.S. Yang, D.S. Gu and B.K. Choi: Mechatronics Vol. 19(5) (2009) pp.680-689.

[9] A. Rakotomamonjy, S. Canu, Frame: J. of Mach. Learn. Res. Vol. 6(9) (2005) pp.1485-1515.

[10] J. Gao, F. Chen, D. Shi, On the construction of support wavelet network, IEEE Int. Conf. on Syst., Man, and Cyber., 2004, pp.3204-3207.

[11] L. Zhang, W. Zhou, L. Jiao: IEEE Trans. on Syst., Man, and Cybern. B: Cybernetics. Vol. 34(1) (2004) pp.34-39.

DOI: 10.1109/tsmcb.2003.811113

[12] G. Chen, G. Dudek, Auto-correlation wavelet support vector machine and its application to regression, in Proc. of the 2nd Canadian Conf. on Comp. and Rob. Vis. (CRV'05), (2005).

DOI: 10.1109/crv.2005.19

[13] Z. Yu, Y. Cai, Least squares wavelet support vector machines for nonlinear system identification. LNCS 3497 (2005) 436-441.

DOI: 10.1007/11427445_71

[14] V.N. Vapnik, The Nature of Statistical Learning Theory, Springer, New York, (1999).

[15] A. Rakotomamonjy, S. Canu: J. of Mach. Learn. Res. Vol. 6(9) (2005) pp.1485-1515.

[16] A. Widodo, B.S. Yang: Exp. Syst. with Appl. Vol. 33(1) (2007) pp.241-250.

[17] C. W. Hsu, C. J. Lin: IEEE Trans. on Neu. Net. Vol. 13(2) (2002) pp.415-425.

[18] V. N. Vapnik, Estimation Dependences Based on Empirical Data, Springer Verlag, Berlin, (1982).