Uncertainty extraction based multi-fault diagnosis of rotating machinery

Ma, Sai, Li, Shunming and Xiong, Yeping (2016) Uncertainty extraction based multi-fault diagnosis of rotating machinery. Journal of Vibroengineering, 18 (1), 139-150.

Abstract

Feature extraction has always been a significant research topic for in-situ fault diagnosis applications. In this research, measurement uncertainty of vibration signal is defined and extracted as a pre-processing step for statistical feature calculation. An Empirical Mode Decomposition (EMD) detrending method combined with hurst exponent criterion is applied to extract uncertainty. Decision tree and Least Square Support Vector Machine (LS-SVM) algorithms are introduced as statistical feature selector and classifier respectively. Misalignment, rub-impact, pedestal looseness as well as eccentricity faults are set on experimental rig in sequence for data collecting and to test the proposed method. As the diagnosis accuracy shows, the extracted uncertain components are more sensitive to rotor faults compared to original vibration signal. HE-EMD (Hurst Exponent-Empirical Mode Decomposition) is proved a rational tool to pre-process vibration signal for an enhanced diagnosis ability. This paper shows effectiveness of a multi-fault diagnosis method with uncertainty components as state indicator and thus provides new approaches for condition monitoring of rotating machinery.

This record has no associated files available for download.

Contributors

Author: Yeping Xiong

Download statistics