Cooperative state of charge and state of health estimation for lithium-ion battery packs: An adaptively tuned dual extended Kalman filter enhanced by gated recurrent unit

To enhance the accuracy of state of charge (SOC) and state of health (SOH) estimation for lithium-ion battery packs under wide temperature range, dynamic loads, and aging processes, a dual-branch gated recurrent unit (GRU)-driven dual-time scale dual extended Kalman filter (DEKF) collaborative estimation framework is proposed. An innovative temperature-adaptive equivalent circuit model is designed to characterize the internal polarization characteristics of battery packs at different temperatures. Battery current, voltage, and temperature time-series features are fed into a dual-branch GRU network, where GRU-A tunes noise covariance in real time and GRU-B triggers the slow time-scale update. Mean squared error (MSE) and Weighted Binary Cross-Entropy are adopted as loss functions for the two tasks respectively. The dual-time-scale DEKF enables real-time tracking of SOC at the fast time scale, and updates ohmic internal resistance and rated capacity to calculate SOH when the triggering conditions are met at the slow time scale. Validation is performed on a public battery dataset from McMaster University. Experimental results show that under the LA92 dynamic driving cycle at room temperature, SOC estimation achieves MAE of 0.24% and RMSE of 0.25%, while SOH obtains MAE of 0.56% and RMSE of 0.76%. At 40 °C and 0 °C, the MAEs of SOC estimation are 0.31% and 0.42%, and the MAEs of SOH estimation are 0.61% and 0.42%, respectively. The proposed model surpasses conventional DEKF and GRU methods with good robustness and accuracy. Benefiting from the combination of the temperature-adaptive equivalent circuit model, dual-branch GRU network, and dual-time-scale DEKF, the presented framework achieves low estimation errors across broad temperature and dynamic operating conditions, providing an accurate and reliable solution for joint SOC-SOH estimation of lithium-ion battery packs.

Authors

Institutions

Publication Details

Journal
Journal of Energy Storage
Published
2026-10-09
DOI
https://doi.org/10.1016/j.est.2026.125055
Primary Topic
Advanced Battery Technologies Research
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Cooperative state of charge and state of health estimation for lithium-ion battery packs: An adaptively tuned dual extended Kalman filter enhanced by gated recurrent unit

Yuhang Liu, Zhen Gao, Qingshan Xu, Li Cai et al.
Journal of Energy Storage
Advanced Battery Technologies Research
article

Cooperative state of charge and state of health estimation for lithium-ion battery packs: An adaptively tuned dual extended Kalman filter enhanced by gated recurrent unit

Yuhang Liu, Zhen Gao, Qingshan Xu, Li Cai, Hao Chen, Siyu He, Leida Li, Ni'na Dai
article en

Abstract

To enhance the accuracy of state of charge (SOC) and state of health (SOH) estimation for lithium-ion battery packs under wide temperature range, dynamic loads, and aging processes, a dual-branch gated recurrent unit (GRU)-driven dual-time scale dual extended Kalman filter (DEKF) collaborative estimation framework is proposed. An innovative temperature-adaptive equivalent circuit model is designed to characterize the internal polarization characteristics of battery packs at different temperatures. Battery current, voltage, and temperature time-series features are fed into a dual-branch GRU network, where GRU-A tunes noise covariance in real time and GRU-B triggers the slow time-scale update. Mean squared error (MSE) and Weighted Binary Cross-Entropy are adopted as loss functions for the two tasks respectively. The dual-time-scale DEKF enables real-time tracking of SOC at the fast time scale, and updates ohmic internal resistance and rated capacity to calculate SOH when the triggering conditions are met at the slow time scale. Validation is performed on a public battery dataset from McMaster University. Experimental results show that under the LA92 dynamic driving cycle at room temperature, SOC estimation achieves MAE of 0.24% and RMSE of 0.25%, while SOH obtains MAE of 0.56% and RMSE of 0.76%. At 40 °C and 0 °C, the MAEs of SOC estimation are 0.31% and 0.42%, and the MAEs of SOH estimation are 0.61% and 0.42%, respectively. The proposed model surpasses conventional DEKF and GRU methods with good robustness and accuracy. Benefiting from the combination of the temperature-adaptive equivalent circuit model, dual-branch GRU network, and dual-time-scale DEKF, the presented framework achieves low estimation errors across broad temperature and dynamic operating conditions, providing an accurate and reliable solution for joint SOC-SOH estimation of lithium-ion battery packs.

Journal of Energy StorageVol. 182
Xidian University (CN), China University of Mining and Technology (CN), China Automotive Battery Research Institute (CN), Chongqing Three Gorges University (CN), Southeast University (CN), McMaster University (CA)
Openalex Percentile: Top 21%
Advanced Battery Technologies Research
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.