Fault diagnosis for bearing based on Mahalanobis-Taguchi system

This paper presents a method for fault diagnosis based on Mahalanobis-Taguchi system (MTS), which is applied to practical fault diagnosis for rolling element bearing. Firstly, this method utilizes time/frequency domain analysis for feature extraction from the vibration data. Then, a computational scheme based on Mahalanobis distance (MD) is used for fault clustering. In addition, Taguchi methods are employed to reduce the redundant characteristics and extract the principal components. Its performance and effectiveness has also been validated via rolling element bearing test-bed. Generally, the MTS-based diagnosis method has two benefits: (1) It is a reliable real-time multivariate analysis; (2) It offers a systematic way to determine the principal characteristics.