This paper studies the roller bearing fault diagnosis method with harmonic wavelet packet and Decision Tree-Support Vector Machine (DT-SVM). The harmonic wavelet packet possesses better performances for its box-shaped spectrum and unlimited subdivision compared with the conventional time-frequency feature exaction method. Firstly, the proposed method decomposes the roller bearing vibration signal with harmonic wavelet packet and extracts the feature energy with coefficients of each spectrum. After the feature energy is normalized, feature vector are available. Based on multi-level binary tree, this paper designs the multi-classification SVM model due to its superior nonlinear mapping capability. Three two-classifications are incorporated to diagnosis the roller bearing faults. Finally, the proposed method is illustrated with the vibration data from the roller bearing stand of electric engineering lab in case western reserve university. Experimental results illustrate the higher accuracy of the proposed method compared with conventional method.
[1]
Chih-Jen Lin,et al.
A Comparison of Methods for Multi-class Support Vector Machines
,
2015
.
[2]
Peng Fu-qiang.
Generalized Demodulation Method Based on Multi-scale Chirplet and Sparse Signal Decomposition and Its Application to Roller Bearing Fault Diagnosis
,
2010
.
[3]
Hong Duan,et al.
Harmonic wavelet analysis of modulated tunable diode laser absorption spectroscopy signals.
,
2009,
Applied optics.
[4]
Chih-Jen Lin,et al.
A comparison of methods for multiclass support vector machines
,
2002,
IEEE Trans. Neural Networks.
[5]
Zhou Huaichun.
State Identification of Boiler Combustion Flame Images Based on Gray Entropy Multiple Thresholding and Support Vector Machine
,
2013
.
[6]
D. Newland.
Wavelet Analysis of Vibration: Part 1—Theory
,
1994
.