MoEKAN: Multi-Scale Transformer-Based Gating KAN Experts Network for Time Series Forecasting
Sensors · 29.11.2025 · 25 (23) · 7287
DOI: 10.3390/s25237287 ↗Kurz gefasst
Fachartikel in der Zeitschrift Sensors, online erschienen am 29.11.2025. Verfasst von Donghyun Kim und 3 weiteren Autor:innen. Beteiligte Einrichtung: Department of Medical Artificial Intelligent, Eul-Ji University, Seongnam-si 13135, Republic of Korea. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 39 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Ministry of Trade, Industry and Energy, Korea Institute of Energy Technology Evaluation and Planning, National Information Society Agency.
Abstract (Originalsprache)
An on-load tap changer (OLTC) is a critical component of power transformers, and the vibration signals generated during its operation provide valuable information for forecasting equipment conditions and anomalies. In this study, we propose a novel mixture-of-experts-based Kolmogorov–Arnold network (KAN) model, referred to as MoEKAN, to enhance the accuracy of time series forecasting for the vibration signals of the OLTC. The proposed MoEKAN incorporates reversible instance normalization (RevIN) to flexibly adapt to changes in data distribution and employs a transformer-based gating mechanism to dynamically integrate forecasts from various KAN expert models. In addition, multi-scale signal processing is performed to effectively capture the complex periodicity and patterns present in the vibration data. Experiments using real OLTC operational data demonstrate that the MoEKAN model achieves superior forecasting performance, recording forecasting errors with MSE, MAE, and MAPE values of 133.4579, 7.2801, and 4.4272, respectively, outperforming all comparison models. These results validate the practicality and contribution of the proposed model and confirm its potential as a highly reliable diagnostic tool for condition monitoring and predictive maintenance of OLTCs.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Medical Artificial Intelligent, Eul-Ji University, Seongnam-si 13135, Republic of Korea.
- Korea Electrotechnology Research Institute, Ansan-si 15588, Republic of Korea.
- Department of Radiological Science, Eul-Ji University, Seongnam-si 13135, Republic of Korea.
ORCID-Profile: Donghyun Kim, Jimyung Kang
Förderung
- Ministry of Trade, Industry and Energy
- Korea Institute of Energy Technology Evaluation and Planning
- National Information Society Agency
Zitieren
APA
Kim, D., Kang, J., Hwang, H., & Kim, H. (2025). MoEKAN: Multi-Scale Transformer-Based Gating KAN Experts Network for Time Series Forecasting. Sensors, 25(23), 7287. https://doi.org/10.3390/s25237287
Vancouver
Kim D, Kang J, Hwang H, Kim H. MoEKAN: Multi-Scale Transformer-Based Gating KAN Experts Network for Time Series Forecasting. Sensors. 2025;25(23):7287. doi:10.3390/s25237287
BibTeX
@article{Kim2025,
author = {Kim, Donghyun and Kang, Jimyung and Hwang, Hoseong and Kim, Hochul},
title = {MoEKAN: Multi-Scale Transformer-Based Gating KAN Experts Network for Time Series Forecasting},
journal = {Sensors},
year = {2025},
volume = {25},
number = {23},
pages = {7287},
doi = {10.3390/s25237287}
}
EUVIMED