Least Squares Approach to Improve Standardized Estimation of OCV and Internal Resistance in Batteries

This paper focuses on standardized approaches for estimating battery open-circuit voltage (OCV) and internal resistance, which rely on low-fidelity models and simple estimation techniques, yet offer repeatable and widely used results in practical applications. In particular, we examine the common method of computing internal resistance as the ratio of the voltage response to an applied current excitation pulse. When applied to batteries, this method implicitly assumes that the OCV remains constant during the pulse, which can introduce significant estimation errors when this assumption is violated. To address this limitation, we propose a novel observation model that explicitly accounts for OCV variation during the excitation window. The model introduces a single parameter representing the OCV-SOC gradient, which is estimated using a linear least-squares approach from the current excitation and voltage response data. This leads to a closed-form solution that requires no prior knowledge of battery-specific parameters such as capacity or equivalent circuit elements. The proposed approach retains the simplicity and short excitation duration of standardized methods, yet yields significantly improved accuracy in estimating both OCV and resistance. Simulation and experimental results show that the method reduces voltage prediction error by up to 85% and estimates internal resistance within 0.04% of the true value under standardized pulsed conditions. Additional validation under dynamic load scenarios further demonstrates the robustness and accuracy of the approach, making it well-suited for embedded applications and standardized diagnostic protocols.