Title: Deep learning rotor fault detection using the Gramian angular field and the Markov transition field encoding approaches

Authors: Aroui Tarek; Marmouch Sameh

Addresses: Université de Sousse, Ecole Nationale d'Ingénieurs de Sousse, 4054, Sousse, Tunisia; Université de Sfax, Ecole Nationale d'ingénieurs de Sfax, Laboratoire des Sciences et Techniques de l'Automatiques et de l'informatique industrielle (LabSTA), 3038, Sfax, Tunisia ' Université de Sousse, Ecole Nationale d'Ingénieurs de Sousse, 4054, Sousse, Tunisia; Université de Sfax, Ecole Nationale d'ingénieurs de Sfax, Laboratoire des Sciences et Techniques de l'Automatiques et de l'informatique industrielle (LabSTA), 3038, Sfax, Tunisia

Abstract: Using deep learning techniques for diagnostic purposes has garnered considerable interest and demonstrated encouraging outcomes across diverse fields. Convolutional neural networks (CNN) are extensively employed in image-based diagnostic tasks. This paper proposes an approach that transforms the temporal signal of the stator current into visual representations. The combined utilisation of the Gramian angular field (GAF) and the Markov transition field (MTF) allows for a comprehensive analysis of stator current data. This combined approach is associated with two deep learning models (VGG19 and RESNET50) for rotor broken bars detection. We have used an experimental database acquired under varied load conditions and with different types of rotor faults to demonstrate the reliability of our approach. The results obtained demonstrate the efficacy of the proposed strategy.

Keywords: rotor faults; diagnosis; GAF; Gramian angular field; Markov transition field; VGG19; RESNET50.

DOI: 10.1504/IJAAC.2025.145924

International Journal of Automation and Control, 2025 Vol.19 No.3, pp.350 - 369

Received: 14 Aug 2023
Accepted: 03 Jul 2024
Published online: 30 Apr 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article