An Adaptive Robust Square Root Unscented Kalman Filter for State of Charge Estimation for Lithium-ion Batteries

Accurate estimation of State of Charge (SOC) is essential for the efficiency, safety, and durability of battery-powered devices, playing a vital role in Battery Management Systems (BMS). This paper introduces an innovative method combining the Adaptive Robust Square Root Unscented Kalman Filter (ARSRUKF) with Recursive Least Squares (RLS) to enhance SOC estimation accuracy and robustness. By maintaining semi-positive definite covariance matrices, the proposed method ensures numerical stability, avoiding issues commonly encountered in traditional techniques. A key feature of the ARSRUKF is its direct computation of the square roots of covariance matrices, preserving their symmetry and positive definiteness while increasing computational efficiency. Unlike conventional filters, such as the Extended Kalman Filter (EKF) and Unscented Kalman Filter (UKF), the ARSRUKF operates effectively without prior knowledge of noise statistics, accommodating non-Gaussian noise or uncertain noise characteristics in real-world scenarios.To further improve performance, an Adaptive Neuro-Fuzzy Inference System (ANFIS) dynamically tunes noise covariances in real-time, adapting to changes in operating conditions like temperature variations, battery aging, and load shifts. Extensive experimental results highlight the superior performance of the ARSRUKF compared to the EKF and UKF, particularly in conditions with unknown or varying noise statistics. This approach demonstrates significant advancements in SOC estimation accuracy, stability, and consistency. The proposed method has broad potential applications in electric vehicles, renewable energy storage, and portable electronics, offering a robust and efficient solution for advanced battery systems.

کلیدواژه ها:

State of charge ، adaptive robust square root unscented Kalman filter ، lithium-ion batteries