ABSTRACT
Combining non-linear factors such as current, temperature, coulombic efficiency etc., a battery capacity correction equation for State of Charge (SOC) is established. The battery state space model is built by using the ampere-hour counting method and composite electrochemical principles. Based on the adaptability of a non-Gaussian and non-linear system, a Particle Filter (PF) algorithm is used to estimate the SOC of the battery. The simulation is performed under random dynamic charge/discharge condition. The result shows a good agreement between the experimental data processed by the PF algorithm and the raw value. Experiments also show that the estimation accuracy adopted with the PF method is 0.5% compared with that of the ampere-hour method, which may apply in a battery management system for runtime SOC estimation.
