Bearing Failure Analysis in Centrifugal Pumps Utilizing Adaptive Convolutional Neural Networks (CNNs)
Open Access DepositedAbstract of Praxis Bearing Failure Analysis in Centrifugal Pumps UtilizingAdaptive Convolutional Neural Networks (CNNs) An important consideration in the context of industrial processes is bearing failures in centrifugal pumps, which can result in abrupt downtimes and additional maintenance costs. Traditional predictive maintenance practices do not have the requisite flexibility capable of reacting to changes in harsh engineering environments. The novelty of this research incorporates early speed and vibration input into an Adaptive Convolutional Neural Network (CNN) architecture to recognize faults under variable speed conditions. The trained model is able to recognize features of normal wear and the beginnings of failure in bearings. At the same time, it is able to adapt its predictions depending on operational conditions. Performance was verified by running the data through both the Static and Adaptive models and comparing them against each other. The Adaptive model provided 13.8% higher fault detection accuracy and 22% higher F1-score. All of the performance gains were statistically significant (p < 0.05). These outcomes showed that real-time speed context integration can provide reliable and robust diagnostics to rotating machinery. It is possible to implement a shift in current preventative maintenance strategies with the help of this research by detecting bearing faults sooner. This will reduce unplanned downtime, extend equipment life, and lower operational costs in centrifugal pump systems.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.