Advancing SOC estimation in LiFePO4 batteries: Enhanced dQ/dV curve and short-pulse methods

Accurate state-of-charge (SOC) estimation for lithium iron phosphate (${\mathrm{LiFePO}}_4$) batteries remains challenging due to their inherently flat open-circuit voltage (OCV)–SOC characteristics, which impair observability for conventional voltage-based and equivalent circuit model (ECM) methods. To address this limitation, we propose a DQV-based SOC estimation framework that uses short-duration current pulses to extract informative voltage features. Complete DQV–SOC reference curves are constructed offline across multiple C-rates ($\pm$ 1/30C, $\pm$ 0.2C, $\pm$ 0.5C, $\pm$ 1C, and $\pm$ 2C). During operation, voltage responses from brief current pulses are processed via exponential fitting to generate smooth, noise-resilient DQV segments. These segments are fused with the reference data within an Unscented Kalman Filter (UKF), enabling closed-loop SOC estimation with low computational overhead. Experimental results highlight the significant influence of C-rates on the DQV-based SOC estimator. We observe that pulse currents significantly enhance SOC estimation convergence across the full SOC range [0, 1]. However, employing a single C-rate pulse may not ensure robustness across diverse SOC ranges, emphasizing the importance of carefully selecting C-rates to achieve SOC estimation convergence throughout the entire SOC range of [0, 1]. This research contributes to advancing reliable management practices for ${\mathrm{LiFePO}}_4$ batteries in electric vehicles.