Real-time and accurate estimation of the state of health (SOH) is of great importance for the maintenance of Li-ion battery in electric vehicles. Li-ion batteries is a complex electrochemical dynamic system, so traditional modeling methods are difficult to achieve online estimation of SOH. Therefore, starting from real-time assessment of SOH of battery, this paper chooses indicators related to the state of battery health and establishes a SOH prediction model based on incremental learning. A HI-DD algorithm based on sliding window technology is proposed for considering the time-consuming problem of online learning in incremental learning, which can detect whether concept drift occurs, thus guiding and determining the updating position of the model. Furthermore, the model updating strategy based on HI-DD and AdaBoost.RT is designed to improve the online learning performance and prediction accuracy of the model. Finally, the proposed method is verified by using the experimental data of battery aging provided by CALCE. The results show that the HI-DD-AdaBoost.RT prediction algorithm based on incremental learning has strong updating ability of online and high accuracy of prediction, which can meet the actual needs of SOH online prediction.