Abstract: State-of-charge (SOC) is a very important part of lithium-ion battery prediction and health management. As the SOC of the lithium battery cannot be measured directly, this paper presents a method of estimating the SOC of the lithium-ion battery by the random forest regression. Firstly, a random forest regression model is constructed. The battery current, battery voltage and battery temperature are used as the training input of the model, and the corresponding SOC is used as the training output of the model. And then, the random forest algorithm for model training is conducted. Finally, the training model is applied to the battery SOC estimation. The results show that by the random forest regression algorithm, the maximum estimation error of the lithium-ion battery charge is 0.02,and the RMSE is 0.003204.This method can effectively estimate the lithium-ion battery SOC and has high estimation accuracy. This model may provide a reference for the model construction of future battery charge estimation system.

Key words: lithium-ion battery, random forest regression, state-of-charge (SOC) estimation