Gaussian Process Regression-Driven SOC Estimation for High-Gain DC-DC Converters in EV Applications

State of Charge (SOC) estimation is a critical component in optimizing the performance and longevity of battery management systems (BMS), especially in applications involving electric vehicles (EVs) and renewable energy storage systems. Accurate SOC estimation ensures efficient power utilization, battery safety, and system reliability. This paper investigates two widely recognized techniques for SOC estimation: The Extended Kalman Filter (EKF) and Gaussian Process Regression (GPR). To achieve a high output voltage level for accurate measurement, a high-gain DC-DC converter is integrated into the system. Both EKF and GPR are applied to estimate the SOC based on the voltage and current measurements from the system. The EKF, a recursive, model-based filtering technique, is compared with the non-parametric, data-driven GPR method, which can model complex nonlinearities and handle noise more effectively. The results from both techniques are analyzed in terms of their accuracy and robustness under varying conditions. Based on the comparative analysis, the paper highlights the superior performance of one technique over the other in terms of estimation precision, particularly in environments with significant measurement noise and nonlinearities. The findings provide valuable insights into the practical application of advanced SOC estimation methods in modern BMS, suggesting a path toward more reliable and efficient battery management solutions.