Physics‑Informed LSTM and EKF Fusion for Robust Battery SoC Estimation

Abstract

Reliable state‑of‑charge estimation remains difficult in the presence of long‑term ageing, open‑circuit‑voltage hysteresis, and variable operating profiles. We investigate a hybrid estimator that marries a first‑principles equivalent‑circuit model with sequence learners, coupled through a residual‑aware fusion rule. Using a minute‑resolution ageing dataset from a lithium‑titanate (LTO) cell (236,282 samples across 2500 cycles), we identify phase‑aware OCV‑to‑SoC lookup tables for charge and discharge, deploy an extended Kalman filter with soft half‑cycle anchoring, and train LSTM, 1D‑CNN, and Transformer baselines with physics‑informed regularization promoting Coulomb consistency and concordance between OCV and terminal voltage. The phase‑aware EKF attains MAE 0.04535 and RMSE 0.05057 on cycle‑wise averages. A stitched LSTM yields MAE 0.03023 and RMSE 0.05192. Residual‑weighted fusion of EKF and LSTM produces MAE 0.03049 and RMSE 0.03991, which represents a 21% reduction relative to EKF while preserving the LSTM’s low bias. A coarse parameter sweep over the circuit model confirms expected early‑life behavior, namely lower resistance and a longer polarization time constant. At the window level, a physics‑informed LSTM achieves MAE 0.0184. The fusion is model‑agnostic, requires no retraining of the base estimators, and adds negligible computational overhead. We release phase‑aware OCV tables, a trained LSTM, and configuration files for seamless one‑command reproduction.

Keywords

• Extended Kalman Filter
• Hybrid State-of-Charge Estimation
• Lithium-Ion Battery Aging
• Physics-Informed Neural Networks
• Residual-Weighted Fusion.

Dayanıklı Pil SoC Tahmini için Fizik Bilgisine Dayalı LSTM ve EKF Füzyonu

Abstract

Uzun vadeli yaşlanma, açık devre voltajı histerezisi ve değişken çalışma profilleri söz konusu olduğunda, güvenilir şarj durumu tahmini yapmak zor olmaya devam etmektedir. Biz, kalıntı farkında füzyon kuralı ile birleştirilen, ilk ilkelere dayalı eşdeğer devre modelini sekans öğrenicilerle birleştiren bir hibrit tahminciyi araştırıyoruz. Lityum titanat (LTO) hücresinden elde edilen dakika çözünürlüklü yaşlanma veri setini (2500 döngüde 236.282 örnek) kullanarak, şarj ve deşarj için faz farkında OCV-SoC arama tabloları belirliyor, yumuşak yarım döngü sabitleme ile genişletilmiş Kalman filtresi uyguluyor ve LSTM, 1D-CNN ve Transformer temel modellerini, OCV ve terminal voltajı arasında Coulomb tutarlılığını ve uyumu destekleyen fizik bilgisine dayalı düzenleme ile eğitiyoruz. Faz farkında EKF, döngü bazında ortalamalarda MAE 0,04535 ve RMSE 0,05057 değerlerine ulaşıyor. Birleştirilmiş LSTM, MAE 0,03023 ve RMSE 0,05192 değerleri veriyor. EKF ve LSTM'nin kalıntı ağırlıklı füzyonu, MAE 0,03049 ve RMSE 0,03991 değerleri verir; bu, LSTM'nin düşük önyargısını korurken EKF'ye göre %21'lik bir azalma anlamına gelir. Devre modeli üzerinde yapılan kaba parametre taraması, beklenen erken yaşam davranışını, yani daha düşük direnç ve daha uzun polarizasyon zaman sabitini doğrulamaktadır. Pencere düzeyinde, fizik bilgisine dayalı LSTM, MAE 0,0184 değerine ulaşmaktadır. Füzyon, modelden bağımsızdır, temel tahmincilerin yeniden eğitilmesini gerektirmez ve ihmal edilebilir düzeyde hesaplama yükü ekler. Kesintisiz tek komutlu yeniden üretim için faz farkında OCV tabloları, eğitilmiş LSTM ve yapılandırma dosyalarını yayınlıyoruz.

Keywords

• Genişletilmiş Kalman Filtresi
• Hibrit Şarj Durumu Tahmini
• Lityum İyon Pil Yaşlanması
• Fizik Bilgisine Dayalı Sinir Ağları
• Artık Ağırlıklı Füzyon.

References

1. International Energy Agency, “Global EV Outlook 2024,” 2024. [Online]. Available: https://www.iea.org/reports/global-ev-outlook-2024
2. GL. Plett, “Extended Kalman filtering for battery management systems of LiPB-based HEV battery packs: Part 3. State and parameter estimation,” J Power Sources, vol. 134, no. 2, pp. 277–292, 2004, doi: https://doi.org/ 10.1016/j.jpowsour.2004.02.033.
3. H. Fang, Y. Wang, Z. Sahinoglu, T. Wada, and S. Hara, “State of charge estimation for lithium-ion batteries: An adaptive approach,” Control Eng Pract, vol. 25, pp. 45–54, 2014, doi: https://doi.org/10.1016/j.conengprac.2013.12.006.
4. S. Yuan, H. Wu, and C. Yin, “State of Charge Estimation Using the Extended Kalman Filter for Battery Management Systems Based on the ARX Battery Model,” Energies (Basel), vol. 6, no. 1, pp. 444–470, 2013, doi: https://doi.org/ 10.3390/en6010444.
5. D. Sun, X. Yu, C. Zhang, C. Wang, and R. Huang, “State of charge estimation for lithium-ion battery based on an intelligent adaptive unscented Kalman filter,” Int J Energy Res, vol. 44, no. 14, pp. 11199–11218, 2020, doi: https://doi.org/10.1002/er.5690.
6. GL. Plett, “Extended Kalman filtering for battery management systems of LiPB-based HEV battery packs: Part 1: Background,” J Power Sources, vol. 134, no. 2, pp. 262–276, 2004.
7. IC. Dikmen, N. Yildiran, and T. Karadag, “Machine Learning Approaches for Enhancing the SoH Estimation of LTO Batteries,” International Journal of Automotive Science and Technology (IJASTECH), vol. 9, no. 1, pp. 48–59, 2025, doi: https://doi.org/10.30939/ijastech..1522403
8. N. Yildiran, IC. Dikmen, and T. Karadag, “State of Health Estimation of Lithium Titanate Oxide Batteries Through Data-Driven Techniques and Machine Learning,” in 2024 8th International Artificial Intelligence and Data Processing Symposium (IDAP), 2024, 10.1109/IDAP64064.2024.10711165.

9. A. Yunusoglu, D. Le, M. Isik, K. Tiwari, IC. Dikmen, and T. Karadag, “Battery State of Health Estimation Using LLM Framework,” in 2025 26th International Symposium on Quality Electronic Design (ISQED), Apr. 2025, pp. 1–8. doi: https://doi.org /10.1109/ISQED65160.2025.11014446.
10. E. Chemali, PJ. Kollmeyer, M. Preindl, R. Ahmed, and A. Emadi, “Long short-term memory-networks for accurate state of charge estimation of Li-ion batteries,” IEEE Transactions on Industrial Electronics, vol. 65, no. 8, pp. 6577–6587, 2018.
11. HM. Hussein, A. Aghmadi, MS. Abdelrahman, SMSH. Rafin, and O. Mohammed, “A review of battery state of charge estimation and management systems: Models and future prospective,” WIREs Energy and Environment, vol. 13, no. 1, p. e507, 2024, doi: https://doi.org/10.1002/wene.507.
12. H. Yu, H. Lu, Z. Zhang, and L. Yang, “A generic fusion framework integrating deep learning and Kalman filter for state of charge estimation of lithium-ion batteries: Analysis and comparison,” J Power Sources, vol. 623, p. 235493, 2024, doi: https://doi.org/10.1016/j.jpowsour.2024.235493.
13. F. Wang, Z. Zhai, Z. Zhao, Y. Di, and X. Chen, “Physics-informed neural network for lithium-ion battery degradation stable modeling and prognosis,” Nat Commun, vol. 15, no. 1, p. 4332, 2024, doi: https://doi.org/10.1038/s41467-024-48779-z.
14. MS. Çetin, MT. Gençoğlu, and A. Dobrzycki, “Investigation of Charging Technologies for Electric Vehicles,” Turkish Journal of Science and Technology, vol. 19, no. 1, pp. 97–106, 2024, doi: https://doi.org/10.55525/tjst.1399120.
15. IC. Dikmen, YE. Ekici, T. Karadağ, T. Abasov, and SE. Hamamcı, “Electrification in Urban Transport: A Case Study with Real-time Data,” Balkan Journal of Electrical and Computer Engineering, vol. 9, no. 1, pp. 69–77, 2021, doi: https://doi.org/10.17694/bajece.862931.
16. YE. Ekici, IC. Dikmen, M. Nurmuhammed, and T. Karadağ, “A Review on Electric Vehicle Charging Systems and Current Status in Turkey,” International Journal of Automotive Science And Technology, vol. 5, no. 4, pp. 316–330, 2021, doi: https://doi.org/10.30939/ijastech..1011270.
17. D. Le, A. Yunusoglu, K. Tiwari, M. Isik, and IC. Dikmen, “Multimodal llm for intelligent transportation systems,” arXiv preprint arXiv:2412.11683, 2024, doi: https://doi.org/10.48550/arXiv.2412.11683.
18. IC. Dikmen, T. Karadağ, and A. Arı, “2500 Cycle Single Cell Battery Aging Dataset,” 2025, IEEE. doi: https://doi.org/10.21227/0svj-e338.
19. T. Karadağ, T. Abbasov, and M. O. Karadağ, “Measuring, evaluating, and mapping the electromagnetic field levels in Turgut Ozal Medical Center building and environment,” Journal of Turgut Ozal Medical Center, vol. 21, pp. 186–195, 2014, doi: https://doi.org/10.7247/JTOMC.2014.1681.
20. YE. Ekici, O. Akdağ, AA. Aydın, and T. Karadağ, “Review and analysis of real-time big data of electric bus consumption data,” in Ankara International Congress on Scientific Research – VII, Ankara, Türkiye, 02 Aralık 2022, ss.1405-1418.
21. Nurmuhammed M., Akdağ O., Karadağ T., “Modified Archimedes optimization algorithm for global optimization problems: a comparative study,” Neural Comput. Appl., 36, 8007–8038, 2024. https://doi.org/10.1007/s00521-024-09497-1