Electric vehicles are emerging as a sustainable solution to global warming, pollution, and fossil fuel depletion, with lithium-ion batteries serving as the primary energy storage devices. Accurate estimation of the State-of-Charge (SOC), reflecting the remaining usable capacity, has been crucial for safe and efficient operation. Accurate SOC determination is required an accurate battery model combined with robust estimation algorithms.

In this work, lithium-ion battery modeling and parameter estimation have been performed using Electrical Equivalent Circuit Models (ECMs), such as 1-RC and 2-RC models. Minor improvements have been observed with the 2-RC model, so the simpler 1-RC model has been used for further study. A BAK A40 18650 cell has been modeled, and parameters have been estimated using experimental current and voltage data from an Arbin testing system. Recursive Least Squares with an adaptive forgetting factor (AFF-RLS) has been proposed to enhance robustness by dynamically adjusting the forgetting factor.

The estimated parameters have been used for SOC estimation via Modified Coulomb Counting (MCC), with the initial SOC having been determined from a piecewise SOC&ndashOCV relationship to improve accuracy. Model-based filtering techniques, including Kalman Filter (KF) and Extended Kalman Filter (EKF), have also been employed. An Adaptive Dual EKF (ADEKF) has been developed for simultaneous SOC and parameter estimation with dynamic noise adaptation, and it has been validated on Arduino Mega 2560, achieving superior accuracy and robustness compared to conventional methods.

Since ECM parameters vary with operating state, aging, and cycle number, they have directly affected capacity degradation. Cell-level parameters have been extracted from ~417 hours of testing and have been used to design a 48 V, 80 Ah battery pack, with ADEKF estimating pack-level SOC and parameters while considering aging. The pack SOH has been determined based on the faded capacity from the piece-wise capacity&ndashcycle relationship.

Finally, an electro-thermal model has been employed to capture the effect of temperature on battery dynamics, and ADEKF has been used to estimate the core temperature of the cell. Internal resistance, capacitance, and OCV&ndashSOC characteristics have been observed to vary with temperature, influencing SOC estimation and dynamic voltage response. Coupling the two-state thermal model with ECM has enabled accurate SOC estimation, reliable parameter estimation, and robust prediction of battery performance under real-time operating conditions.