Estimation of Lithium-ion Battery State of Charge Using Recursive Least Squares Method

This paper addresses one of the major challenges of SoC estimation in the BMS, which is SoC estimation without the knowledge of battery capacity. The proposed SoC estimation algorithm is based on the Recursive Least Squares (RLS) method. The inputs of the proposed algorithm are voltage, current, the time step as the inputs, and the SOC as the output. The RLS algorithm constructs an input and output vector for each sample, given the voltage of the battery at the time instant. The proposed algorithm predicts the SoC, calculates the error between the predicted and the actual SoC, and then uses a forgetting factor to increase accuracy. The characteristic curve is estimated based on the measurement vector and the covariance matrix of the estimation error. Then, current and change in time inputs as well as the estimated battery capacity from the RLS are used in the Coulomb Counting equation to accurately estimate the SoC of lithium-ion batteries. RLS-SoC is implemented and tested for a generic battery model using MATLAB/Simulink. The results are obtained to show that the RLS-SoC can improve the accuracy of SoC estimation compared to the traditional methods without the knowledge of battery capacity. The algorithm is also computationally efficient and can be integrated with real-time BMS applications.