A new framework for intelligent simultaneous-fault diagnosis of rotating machinery using pairwise-coupled sparse Bayesian extreme learning committee machine

Abstract

This paper proposes a new diagnostic framework, namely, probabilistic committee machine, to diagnose simultaneous-fault in the rotating machinery. The new framework combines a feature extraction method with ensemble empirical mode decomposition and singular value decomposition, multiple pairwise-coupled sparse Bayesian extreme learning machines (PCSBELM), and a parameter optimization algorithm to create an intelligent diagnostic framework. The feature extraction method is employed to find the features of single faults in a simultaneous-fault pattern. Multiple PCSBELM networks are built as different signal committee members, and each member is trained using vibration or sound signals respectively. The individual diagnostic result from each fault detection member is then combined by a new probabilistic ensemble method, which can improve the overall diagnostic accuracy and increase the number of detectable fault as compared to individual classifier acting alone. The effectiveness of the proposed framework is verified by a case study on a gearbox fault detection. Experimental results show the proposed framework is superior to the existing single probabilistic classifier. Moreover, the proposed system can diagnose both single- and simultaneous-faults for the rotating machinery while the framework is trained by single-fault patterns only.