Predicting stator winding short-circuit faults in induction motors using machine learning: a comparative study

Aug 4, 2026

Induction motors are essential in industrial systems and electric vehicles, yet stator winding inter-turn short-circuit faults undermine reliability and drive costly downtime. This study evaluates machine learning fault detection by comparing Artificial Neural Networks (ANN), Long Short-Term Memory (LSTM) networks, and Physics-Informed Neural Networks (PINNs). Using a benchmark dataset of stator current signals, leakage flux, and varying load conditions, the models are trained and assessed across seven fault categories and binary health-versus-fault scenarios. Results show near-perfect binary performance, with LSTM at about 100% accuracy and PINN at 99.32%, both exceeding ANN at 99.01%. The findings support temporal modeling and physics-informed learning for more dependable early fault detection, enabling improved predictive maintenance and reduced operational cost.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages29

Reading Time1 h 13 min

AccessFree to download

Introduction

Fault impact on reliability in industrial and EV applications

Motivation for intelligent fault diagnosis

cdr Predicting stator winding short-circuit faults in induction motors using machine learning: a comparative study | Item Type | Article | | --- | --- | | Authors | Aasar, Tayyaba;Sultan , Kiran;Umer Khan , Adnan;Czanner, Silvester;Abbasi , Ayesha;Ahmad , Aasar | | Citation | Aasar, T. , Sultan , K. , Khan , A. U. , Czanner, S. , Abbasi , A. , & Ahmad , A. (2026) . Predicting stator winding short-circuit faultsin induction motors using machine learning: a comparative study. Discover Computing, 29(1), 103 . [https://doi.org/10.1007/](https://doi.org/10.1007/)[ ](https://doi.org/10.1007/)s10791-026-09984-0 | | DOI | 10.1007/s10791-026-09984-0 | | Publisher | Springer;Discover | | Journal | Discover Computing | | Download date | 2026-03-08 16:57:36 | | Item License | [https://creativecommons.org/licenses/by/4.0/](https://creativecommons.org/licenses/by/4.0/) | | Link to Item | [https://chesterrep.openrepository.com/handle/10034/629883](https://chesterrep.openrepository.com/handle/10034/629883) | Aasar etal. Discover Computing (2026) 29:103 [https://doi.org/10.1007/s10791-026-09984-0](https://doi.org/10.1007/s10791-026-09984-0) Discover Computing RESEARCH Open Access Predicting stator winding short-circuit faults in induction motors using machine learning: a comparative study Tayyaba Aasar1, Kiran Sultan2*, Adnan Umer Khan1, Silvester Czanner2, Ayesha Abbasi1 and Aasar Ahmad1 *Correspondence: Kiran Sultan [k.sultan@chester.ac.uk](k.sultan@chester.ac.uk)[ ](k.sultan@chester.ac.uk)1Department of Electrical Engineering, International Islamic University Islamabad (IIUI), Islamabad, Pakistan 2School of Computer and Engineering Sciences, University of Chester (UoC), Chester, UK Abstract Induction motors play a critical role in industrial operations and electric vehicles, yet their reliability is often compromised by stator winding inter-turn short-circuit (ITSC) faults. To mitigate costly downtimes, this study investigates machine learning-based fault detection methods, comparing Artificial Neural Networks (ANN), Long ShortTerm Memory (LSTM) networks, and Physics-Informed Neural Networks (PINNs) . A benchmark dataset comprising stator current signals, leakage flux, and varying load conditions was employed to train and evaluate the models. While classification across seven fault categories remained challenging, the proposed framework achieved outstanding results in binary classification (healthy vs. faulty) . In particular, the LSTM model attained a near-perfect accuracy of 100%, and the PINN model achieved 99. 32%, both surpassing the baseline ANN performance of 99. 01% . These findings highlight the effectiveness of temporal modeling with LSTM and the added value of incorporating motor physics into PINNs, especially in scenarios with limited data. The results confirm that advanced machine learning models can significantly improve early fault detection, enabling more reliable predictive maintenance and reduced operational costs in industrial systems. Keywords Induction motor, Predictive maintenance, Machine learning, Fault detection, Physics-informed neural networks, Stator winding short-circuit fault 1 Introduction Induction motors play a fundamental role in modern industrial and transportation systems and are extensively used in pumps, fans, compressors, conveyors, and increasingly in electric vehicles (EVs) due to their high efficiency, robust construction, low cost, and ease of control [1, 2]. Despite these advantages, induction motors frequently operate under harsh and dynamic conditions, including fluctuating loads, thermal stress, inverter-fed power supplies, and electromagnetic disturbances. Such conditions accelerate the degradation of both electrical insulation and mechanical components, leading to faults such as bearing damage, rotor bar breakage, eccentricity, and stator winding ITSC faults. Among these, ITSC faults are particularly severe, as they can rapidly evolve into © The Author(s) 202

FAQs

Why are stator winding inter-turn short-circuit (ITSC) faults critical in induction motors?

ITSC faults can rapidly progress into more severe phase-to-phase or ground faults, leading to current imbalance, torque oscillations, excessive heating, and potential irreversible motor failure without early detection.

Which machine learning models are compared for fault detection in the study?

The study compares Artificial Neural Networks (ANN), Long Short-Term Memory (LSTM) networks, and Physics-Informed Neural Networks (PINNs) for detecting stator winding ITSC faults.

What data and conditions are used to train and evaluate the models?

Models are trained and evaluated using a benchmark dataset containing stator current signals, leakage flux, and varying load conditions.

Aug 4, 2026

Upload by Valentina