
[ Article ]
Journal of the KNST - Vol. 8, No. 4, pp.760-766
ISSN: 2635-4926
(Print)
Print publication date 31 Dec 2025
Received 19 Oct 2025
Revised 31 Oct 2025
Accepted 17 Dec 2025
머신러닝을 활용한 배터리 열폭주 전조 이상 탐지 기법
안원기1, * ; 박성순2
Machine Learning-based Method for Detecting Thermal Runaway Precursors in Batteries
Wongi Ahn1, * ; Seongsun Park2
Correspondence to: *Wongi Ahn Naval R&D Center, Hanwha Systems 264-60, Sanho-daero, Gumi-si, Gyeongsangbuk-do, 39370, Republic of Korea Tel: +82-54-460-8874 Fax: +82-54-460-8519 E-mail: wg.ahn@hanwha.com
Ⓒ 2025 Korea Society for Naval Science & Technology
초록
본 연구는 리튬이온 배터리의 정상 운전 로그를 기반으로 LSTM Autoencoder를 학습하고, 열폭주 전조 구간을 합성하여 테스트 데이터로 활용하였다. 모델은 재구성 오차(MSE)를 통해 비정상 패턴을 조기에 탐지하였으며, 해군 운용 환경에서의 배터리 안전성 확보를 위한 조기 경고 시스템으로의 적용 가능성을 확인하였다.
Abstract
This study presents an LSTM Autoencoder model trained on normal operation logs of lithium-ion batteries to detect precursor signs of thermal runaway. Synthetic precursor segments were embedded into the test data to simulate early physical anomalies. The model successfully identified abnormal patterns through elevated reconstruction error (MSE), demonstrating its potential as an early warning system for battery safety in naval applications.
Keywords:
Machine Learning, LSTM Autoencoder, Thermal Runaway, Anomaly Detection, Battery Safety키워드:
머신러닝, LSTM 오토인코더, 열폭주, 이상 탐지, 배터리 안전성Acknowledgments
이 논문은 2025년도 한국해군과학기술학회 동계학술대회 발표 논문임.
References
- Zhang, J., Lee, J., & He, Z. (2020). Early Detection of Thermal Runaway in Lithium-ion Batteries using Autoencoder-based Anomaly Detection. Journal of Power Sources, 478, 228753.
- Kim, D., Lee, J., & Park, S. (2021). Unsupervised Anomaly Detection for Battery Safety Using Autoencoder-based Models. Energies, 14(9), 2456.
- Li, Y., Wang, Z., & Chen, M. (2021). Early Detection of Thermal Runaway in Lithium-ion Batteries Using Machine Learning Techniques. Journal of Power Sources, 512, 230–238.
- Zhang, X., He, H., & Xiong, R. (2021). A Data-driven Approach for Lithium-ion Battery Fault Diagnosis Using LSTM Autoencoder. IEEE Transactions on Industrial Electronics, 68(6), 5394–5404.
- Liu, T., Song, Y., & Zhang, Y. (2022). Thermal Runaway Prediction in Lithium-ion Batteries Based on Sensor Fusion and Deep Learning. Applied Energy, 306, 117984.
- 임병주, 조성훈, 이가람, 최석민, 박창대. (2022). 리튬이온 배터리의 열폭주 이상징후 감지를 위한 측정 변수 특성 분석. 한국수소 및 신에너지학회 논문집, 33(1), 85–94.