A Hybrid Kalman Filtering for State of Charge Estimation of Lithium-Ion Battery Used in Low-Powered Microcontroller: CCM-EKF Approach

Abstract

A new technique that aimed to be flashed into a low-power microcontroller to estimate the state of charge (SOC) of lithium-ion battery is introduced. First, an electrical equivalent model is developed to simulate and model the behaviour of the battery. The battery used in this work is Panasonic NCR18650B and 2RC model is chosen to capture the slow and fast response of the battery. The parameters are identified through the discharging cycle with the help of MATLAB Simulink Design Optimization (SDO). Then, the model is verified through simulations and experiments. Once the model is verified to be within an acceptable range, it is then set to run several drive cycles, including constant current discharge (CCD) and hybrid pulse power characterization (HPPC) test. It is shown that the SOC estimation error using Extended Kalman Filter (EKF) is < \mathbf{1}\%$ in both simulation and experiment. Finally, the algorithm is flashed into an STM chip and then external circuits are used to collect the real-world data. The SOC is then estimated. A hybrid Columb Counting Method and EKF SOC estimation are introduced to suit the limited flash memory of the chip. The on-chip SOC estimation error is < \mathbf{0.5}\%$.