Estimate the Parameters of the Polynomial Open Circuit Voltage (OCV) Function LiFePO4 Battery Using the Genetic Algorithm Method

One important factor that affects the driving range and battery life of an electric vehicle is the state of charge (SOC) estimation accuracy. Since the SOC is a hidden state, we are unable to measure it directly. However, by creating a battery-equivalent circuit model, this state can be estimated using the model-based approach. The SOC's performance is directly impacted by the battery model's accuracy. As a result, a suitable equivalent circuit model needs to be found. The SOC-OCV relationship is one of these factors. This paper focuses on using the genetic algorithm (GA) method to estimate the polynomial open-circuit voltage (OCV) parameters function of the LiFePO4 battery. The proposed method is compared to the least squares method. The results show that the GA method can well estimate the parameters of the polynomial function in various degrees, with root mean square error (RMSE) within 1.7% and maximum absolute error within 11% in high degree.