Innovative ISF Fault Detection in PMSMs Using FFT and Artificial Intelligence Techniques

EasyChair Preprint 15850

Abstract

This paper presents an innovative approach for detecting inter-turn short circuit (ISF) faults in Permanent Magnet Synchronous Machines (PMSMs). By using zerosequence voltage signals, identified as the most effective for ISF detection, Fast Fourier Transform (FFT) analysis is applied to extract relevant signal characteristics. Additionally, the integration of Artificial Intelligence (AI) techniques, such as SVM, KNN, and decision trees, allows for the automation of fault detection and classification, enhancing the accuracy and reliability of machine monitoring. The results show that the SVM and KNN models are particularly effective in fault detection, achieving perfect precision and recall. The combined use of these techniques not only optimizes fault detection efficiency but also enhances overall PMSM performance by reducing failure risks and enabling predictive maintenance. This work represents a significant advancement toward smarter and more responsive maintenance solutions for electrical machines.

Keyphrases: Artificial Intelligence, Fast Fourier Transform FFT, Inter turn short circuit ISF, Synchronous Machines PMSM, permanent magnet, zero sequence voltage

@booklet{EasyChair:15850,
  author    = {Oumayma Salhi and Walid Ben Mabrouk and Bilal Amghar and Chakib Ben Njima},
  title     = {Innovative ISF Fault Detection in PMSMs Using FFT  and Artificial Intelligence Techniques},
  howpublished = {EasyChair Preprint 15850},
  year      = {EasyChair, 2025}}