Citation: Lee, H.; Jeong, H.; Kim, S.;
Kim, S.W. Severity Estimation for
Interturn Short-Circuit and
Demagnetization Faults through
Self-Attention Network. Sensors 2022,
22, 4639. https://doi.org/10.3390/
s22124639
Academic Editors: Athanasios
Rakitzis, Khanh T. P. Nguyen and
Kim Phuc Tran
Received: 31 May 2022
Accepted: 17 June 2022
Published: 20 June 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Article
Severity Estimation for Interturn Short-Circuit and
Demagnetization Faults through Self-Attention Network
Hojin Lee, Hyeyun Jeong, Seongyun Kim and Sang Woo Kim *
Department of Electrical Engineering, Pohang University of Science and Technology, 77 Cheongam-Ro, Nam-Gu,
Pohang 37673, Korea; suvvus@postech.edu (H.L.); jhy90@postech.edu (H.J.); ksy3dmbe3kor@postech.edu (S.K.)
* Correspondence: swkim@postech.edu; Tel.: +82-054-279-2237
Abstract:
This study presents a novel interturn short-circuit fault (ISCF) and demagnetization fault
(DF) diagnosis strategy based on a self-attention-based severity estimation network (SASEN). We
analyze the effects of the ISCF and DF in a permanent-magnet synchronous machine and select
appropriate inputs for estimating the fault severities, i.e., a positive-sequence voltage and current
and negative-sequence voltage and current. The chosen inputs are fed into the SASEN to estimate
fault indicators for quantifying the fault severities of the ISCF and DF. The SASEN comprises an
encoder and decoder based on a self-attention module. The self-attention mechanism enhances the
high-dimensional feature extraction and regression ability of the network by concentrating on specific
sequence representations, thereby supporting the estimation of the fault severities. The proposed
strategy can diagnose a hybrid fault in which the ISCF and DF occur simultaneously and does not
require the exact model and parameters essential for the existing method for estimating the fault
severity. The effectiveness and feasibility of the proposed fault diagnosis strategy are demonstrated
through experimental results based on various fault cases and load torque conditions.
Keywords:
deep learning; demagnetization fault; fault diagnosis; interturn short-circuit fault;
permanent-magnet synchronous machine; self-attention; severity estimation
1. Introduction
Condition monitoring and fault diagnosis are fundamental processes for maintain-
ing the advantages of permanent-magnet synchronous machines (PMSMs) for various
applications. Accurate and preemptive fault diagnoses can reduce economic losses and im-
prove reliability and stability by preventing excessive system downtime and accidents [
1
].
Accordingly, many studies have been conducted on diagnosing interturn short-circuit
faults (ISCFs), demagnetization faults (DFs), bearing faults (BFs), and eccentricity faults
(EFs), all of which frequently occur in PMSMs [
2
]. Among these, ISCFs and DFs directly
reduce the efficiency of the PMSM and increase its operational cost. In addition, owing to
their characteristics, each of these faults can lead to the other and/or increase the other’s
severity [
3
]. Therefore, it is essential to accurately diagnose ISCFs and DFs at an early stage.
The ISCF is one of the most frequently occurring stator winding failures, and results
in a short circuit owing to a breakdown of the insulation between adjacent windings [
4
].
The main causes of ISCFs are mechanical, electrical, and thermal stresses [
5
]. A large
amount of circulating current is generated in the short-circuited winding; this increases
the torque ripple owing to the phase imbalance, reducing the efficiency and performance
of the PMSM, and endangering its safe operation [
6
]. In addition, excessive local heat is
generated, accelerating the insulation breakdown and further increasing the severity of the
ISCF [
7
]. Furthermore, the permanent magnets (PMs) in a rotor can be irreversibly demag-
netized, owing to the reductions in the magnetic coercivity and local inverse magnetic field
according to the large circulating current [
8
]. Therefore, it is essential to diagnose an ISCF
at an early stage before it becomes serious and leads to other failures, such as DFs.
Sensors 2022, 22, 4639. https://doi.org/10.3390/s22124639 https://www.mdpi.com/journal/sensors
