FRACTAL FEATURES BASED ROLLER BEARING FAULT ANALYSIS USING MULTI SUPPORT VECTOR MACHINE
Main Article Content
Abstract
In induction motor, condition monitoring gathers much attention which improves the reliability and requires lesser cost of maintenance. There is significant research space for improvement in algorithms and techniques for analyzing the condition of an induction motor. In induction motor failures distribution, around 40% of the failures is due to roller bearing faults. In this work, roller bearing faults such as outer race faults and inner race faults are analyzed using multi support vector machine. The vibration signals of outer and inner race faults along with normal bearing for various loads are considered for this analysis. The 3 dimensional (3D) images are plotted using the data obtained from the vibration signals. Fractal features like fractal dimension, fractal average, fractal standard deviation and lacunarity are extracted from these images using four types of filters namely sobel, prewitt, roberts and canny. These features are fed as input for multi support vector machine (MSVM) for the identification of different types of roller bearing faults using four types of kernel known as Gaussian, RBF, polynomial and sigmoidal. MSVM is operated in two approaches and are one versus one and one versus all. The performance of the MSVM with two different approaches is compared with other methods like Linear discriminant classifier (LDC), Quadratic discriminant classifier (QDA), Decision tree, K-nearest neighbour classifier. RBF kernel based MSVM with one versus all approach usi
Article Details
References
[1] Chen, F.F., Tang, B.P.; Song, Multi-fault diagnosis study on roller bearing based on multi-kernel support vector machine with chaotic particle swarm optimization, Journal of the International Measurement Confederation, 2014, 47 (1), pp. 576–590.
[2] Jiang, L.; Xuan, J.P.; Shi, T.L, Feature extraction based on semi-supervised kernel Marginal Fisher analysis and its application in bearing fault diagnosis, Mechanical Systems and Signal Processing, 2013, 41 (1-2), pp. 113–126
[3] Jolliffe, I.T, Principal Component Analysis’, Series: Springer Series in Statistics’(Springer: New York, NY, USA, 2010, 2nd ed. )
[4] Borg, I.; Groenen, P, Modern Multidimensional Scaling: Theory and Applications, (Springer: New York, NY, USA, 2005, 2nd ed.), pp. 207–212
[5] Martinez, A.M.; Kak, A.C, PCA versus LDA, IEEE Trans. Pattern Anal. Mach. Intell. 2001, 23, 228–233.
[6] Mahamad, A.K.; Hiyama, T, Development of Artificial Neural Network Based Fault Diagnosis of Induction Motor Bearing’, IEEE International Conference on Power and Energy PECon, Johor Bahru, 2008, pp. 1387-1392.
[2] Jiang, L.; Xuan, J.P.; Shi, T.L, Feature extraction based on semi-supervised kernel Marginal Fisher analysis and its application in bearing fault diagnosis, Mechanical Systems and Signal Processing, 2013, 41 (1-2), pp. 113–126
[3] Jolliffe, I.T, Principal Component Analysis’, Series: Springer Series in Statistics’(Springer: New York, NY, USA, 2010, 2nd ed. )
[4] Borg, I.; Groenen, P, Modern Multidimensional Scaling: Theory and Applications, (Springer: New York, NY, USA, 2005, 2nd ed.), pp. 207–212
[5] Martinez, A.M.; Kak, A.C, PCA versus LDA, IEEE Trans. Pattern Anal. Mach. Intell. 2001, 23, 228–233.
[6] Mahamad, A.K.; Hiyama, T, Development of Artificial Neural Network Based Fault Diagnosis of Induction Motor Bearing’, IEEE International Conference on Power and Energy PECon, Johor Bahru, 2008, pp. 1387-1392.