State of Charge Estimation of a Lithium-Ion Battery in an Electric Vehicle Using the XGBoost Method

Accurate state of charge (SOC) estimation for lithium-ion batteries (LIBs) in electric vehicles (EVs) is significant in the battery management system (BMS), Thus, researchers have extensively investigated the development of SOC estimation methods to enhance their reliability and performance. This chapter presents a battery SOC estimation using the extreme gradient boosting (XGBoost) regression algorithm and compares its performance to support vector regression (SVR) and forward neural network (FNN) methods. The objective of the study is to verify the generalization ability of the (XGBoost) method in estimating the battery SOC under different temperatures and driving conditions. The US06 drive cycle was used for training the models, while LA92, UDDS, and HWFET drive cycles were used for testing. The performance of the XGBoost, FNN, and SVR methods was evaluated using the Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) metrics. Based on the experimental findings, it was observed that the (XGBoost) method achieved the lowest RMSE and MAE values with a small execution time compared to the other two methods when enough measurement data samples were available. In conclusion, the XGBoost algorithm shows good generalization ability for different datasets when enough measurement data samples are available.