Authors: Natalia F. Espinoza-Sepulveda; Jyoti K. Sinha
Addresses: Dynamics Laboratory, Department of Mechanical, Aerospace and Civil Engineering (MACE), The University of Manchester, Manchester, UK ' Dynamics Laboratory, Department of Mechanical, Aerospace and Civil Engineering (MACE), The University of Manchester, Manchester, UK
Abstract: A machine learning (ML) model is developed earlier for the rotating machine faults diagnosis. Experimental vibration data in time domain from a rotating rig are used for this ML development. The ML model is developed at a machine speed with different rotor faults and then this experimental ML model is blindly tested with the vibration data at a different machine speed. In this paper, a finite element (FE) model for the rig is developed to understand the dynamics and to validate both, the developed experimental ML model and the vibration-based parameters used. The validation is conducted first at a machine speed and then the model is tested blindly at a different machine speed. It is generally time consuming and often difficult to simulate all kinds of defects and their different sizes in the experimental rig. Therefore, the mathematical FE model of the experimental rig provides the possibility to further extend the research to different defects and other operational conditions.
Keywords: artificial neural network; ANN; machine learning; finite element model; vibration analysis; fault diagnosis.
International Journal of Hydromechatronics, 2021 Vol.4 No.3, pp.295 - 308
Received: 18 Dec 2020
Accepted: 07 Aug 2021
Published online: 29 Sep 2021 *