Physics-Guided Multi-Stage Gradient Boosting for Battery State of Health Estimation
2nd International Symposium on AI-Driven Engineering Systems, ISADES 2026, Hybrid, Mbale, Uganda, 19 - 20 Haziran 2026, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/isades69945.2026.11608199
- Basıldığı Şehir: Hybrid, Mbale
- Basıldığı Ülke: Uganda
- Anahtar Kelimeler: 2-RC, Catboost, ECM, LightGBM, Lithium-ion battery, State of Health, Voltage Relaxation, Xgboost
- Kocaeli Üniversitesi Adresli: Evet
Özet
Lithium-ion batteries are widely used in various industrial applications today due to their high energy density and long lifespan. However, to ensure these batteries operate safely and reliably throughout their cycle life, it is critical to closely monitor their State of Health (SOH). In modern Battery Management Systems (BMS), there is a growing need for advanced systems capable of striking an optimal balance between computational load and prediction accuracy. Furthermore, most conventional data-driven models inevitably require a huge dataset to train successfully. To address this, we focused on the 2RC electrical equivalent circuit model and we developed a Physics-Guided Fusion-Layered Parallel Hybrid SOH prediction framework using Gradient Boosting-based machine learning algorithms. The developed method estimates the SOH by using 2-RC model parameters extracted from the voltage relaxation period following a discharge pulse, along with temperature and operating conditions, as essential inputs. Experimental results demonstrate that the proposed method exhibits low error rates and high generalizability even in different degradation conditions and various temperatures.