This article from the University of Tehran proposes an intelligent method for detecting novel faults in centrifugal pumps using deep Convolutional Neural Networks (CNNs) and unsupervised methods. The study addresses the challenges in data-driven fault diagnosis of rotating machines, particularly the lack of information about the variety of faults the system may encounter. The authors use a combination of t-SNE method and clustering techniques to detect novel faults. The network is augmented using the new data upon detection. The methodology is tested on a centrifugal pump and shows high accuracy in detecting novel faults.
Publication date: 25 Sep 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2309.12765
