Method of Wheel Out-of-Roundness Detection Based on POVMD and Multinuclear LS-SVM

With the continuous increase of the running mileage and speed of the train, more and more wheels become out-of-roundness due to the collision and friction between the wheels and track. It has great significance to detect wheel polygon in order to ensure the safe operation of trains. The wheel out-of-roundness detection method based on POVMD and multinuclear LS-SVM is investigated by using POVMD algorithm to decompose the vibration signal, and then PSO to get optimal parameters which takes VMD algorithm into consideration. Such a method extracts some features from IMF components. Finally, Gaussian kernel function and directed acyclic graph classification method are chosen to build multinuclear classifier to detect wheel out-of-roundness. The experiment results show that the proposed method is effective to analyze wheel out-of-roundness.

[1]  Vladimir Vapnik,et al.  Universal learning technology : Support vector machines , 2005 .

[2]  Ethem Alpaydin,et al.  Multiple Kernel Learning Algorithms , 2011, J. Mach. Learn. Res..

[3]  James Kennedy,et al.  Particle swarm optimization , 2002, Proceedings of ICNN'95 - International Conference on Neural Networks.

[4]  Wang Hong.  On Multiple Kernel Learning Methods , 2010 .

[5]  Yan-Ning Cai,et al.  On Multiple Kernel Learning Methods: On Multiple Kernel Learning Methods , 2010 .

[6]  Yoram Baram,et al.  Learning by Kernel Polarization , 2005, Neural Computation.