Hybrid Method for Fault Diagnosis and Localization in Induction Motors Based on Physics-Informed Features and an Autoencoder Ensemble

Timely fault diagnosis in induction motors is important for preventing unplanned downtime. This paper proposes a method for detecting and localizing faults in the rotor, bearings, and stator winding using an ensemble of autoencoders trained on healthy-condition data. Each autoencoder analyzes a diagnostic representation constructed from current or vibration signals according to the physical manifestations of the corresponding fault. Fault localization is based on standardized reconstruction errors. The linear predictive coding (LPC) order for the rotor and bearing representations is selected using the relative variability in coefficients calculated from healthy data. The method was evaluated on an experimental dataset covering three supply frequencies and three load levels. In the main experiment, mean classification accuracy across ten runs was 99.48±0.09%. Performance varied across the tested reference-motor and operating-condition configurations. The results demonstrate fault localization for the investigated conditions while identifying limitations on transfer to other motors and operating conditions.

Authors

Institutions

Publication Details

Journal
Machines
Published
2026-11-19
DOI
https://doi.org/10.3390/machines14111381
Primary Topic
Machine Fault Diagnosis Techniques
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Hybrid Method for Fault Diagnosis and Localization in Induction Motors Based on Physics-Informed Features and an Autoencoder Ensemble

Olga L. Ibryaeva, Victoria A. Eremeeva, Dmitry Galyshev, Alexander Shestakov
Machines
Machine Fault Diagnosis Techniques
article

Hybrid Method for Fault Diagnosis and Localization in Induction Motors Based on Physics-Informed Features and an Autoencoder Ensemble

Olga L. Ibryaeva, Victoria A. Eremeeva, Dmitry Galyshev, Alexander Shestakov
article en

Abstract

Timely fault diagnosis in induction motors is important for preventing unplanned downtime. This paper proposes a method for detecting and localizing faults in the rotor, bearings, and stator winding using an ensemble of autoencoders trained on healthy-condition data. Each autoencoder analyzes a diagnostic representation constructed from current or vibration signals according to the physical manifestations of the corresponding fault. Fault localization is based on standardized reconstruction errors. The linear predictive coding (LPC) order for the rotor and bearing representations is selected using the relative variability in coefficients calculated from healthy data. The method was evaluated on an experimental dataset covering three supply frequencies and three load levels. In the main experiment, mean classification accuracy across ten runs was 99.48±0.09%. Performance varied across the tested reference-motor and operating-condition configurations. The results demonstrate fault localization for the investigated conditions while identifying limitations on transfer to other motors and operating conditions.

MachinesVol. 14(11)
South Ural State University (RU)
Machine Fault Diagnosis Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon