Authors : Eslam Sayed, Ahmed A. Abdelsamee and Nouby M. Ghazaly
Abstract: The different utilities of centrifugal pumps made the potential for fault occurrence inevitable thus early fault diagnosis is essential for such machines to prevent further losses in different demands. In this study, a vibration-based condition monitoring with the development of the Artificial Neural Network (ANN) model for fault classification and detection. The Multilayer perceptron network with the back-propagation algorithm model is that the most ordinarily used network nowadays. The neural network ability to internally learn from examples makes them more engaging and exciting in the data mining scientific field, rather than following a collection of rules such that by human consultants. This paper deals with the evaluation and development of the ANN model for fault recognition in a centrifugal pumping system with two faults simulated which were seal and particle impurities hitting the impeller. Data required to feed the network extracted from the time-domain vibration raw signal. Results showed great potential for using ANN as a fault diagnosis; the recognition rate of the network was 0.958.
Eslam Sayed, Ahmed A. Abdelsamee and Nouby M. Ghazaly, 2020. Classification Model using Neural Network for Centrifugal Pump Fault Detection. International Journal of Signal System Control and Engineering Application, 13: 120-126.