Title :
A method of multi-class faults classification based-on Mahalanobis-Taguchi system using vibration signals
Author :
Ren, Jiangtao ; Xing, Xiaochen ; Cai, Yuanwen ; Chen, Jing
Author_Institution :
Co. of Postgrad. Manage., Acad. of Equip. Command & Technol., Beijing, China
Abstract :
In this paper, an improved Mahalanobis-Taguchi system based fault diagnosis scheme is presented, vibration signals are used as the signal resource. Mahalanobis-Taguchi System is utilized for fault clustering method in order to classify faults into different categories, Lipschitz Exponents are used to extract characteristic vectors. Firstly, the procedure of implementing Mahalanobis-Taguchi System is introduced, a multi-class faults classification method is proposed based-on improved MTS. Secondly, the mathematical characterization of singularities with Lipschitz exponents is explained, and characteristic vector is obtained based-on Lipschitz exponents. Lastly, the method is validated by rotating machinery seeded-fault-test vibration signals, results showed that this method performed well.
Keywords :
Taguchi methods; fault diagnosis; signal classification; Lipschitz exponents; Mahalanobis-Taguchi system; fault clustering method; fault diagnosis scheme; mathematical characterization; multiclass faults classification; rotating machinery seeded-fault-test vibration signals; Circuit faults; Continuous wavelet transforms; Fault diagnosis; Vibrations; Wavelet analysis; Lipschitz Exponents; Mahalanobis-Taguchi System; Vibration Signals;
Conference_Titel :
Reliability, Maintainability and Safety (ICRMS), 2011 9th International Conference on
Conference_Location :
Guiyang
Print_ISBN :
978-1-61284-667-5
DOI :
10.1109/ICRMS.2011.5979414
