Remaining Useful Life Prediction of Lithium-ion Battery Based on Unscented Kalman Filter and Back propagation Neural Network

Lithium-ion batteries are the key components of many electronic devices and complex systems. In order to decrease the threat of battery failure, as well as to improve the dependability and safety of the system, it is essential to effectively model the batteries performance degradation indicators and accurately predict the remaining useful life (RUL) of the batteries. In this paper, the unscented Kalman filter (UKF) algorithm is fused with back propagation (BP) neural network to increase the prediction accuracy of RUL of lithium-ion batteries. The BP neural network predicts the residual of UKF by auto-regressive form. The UKF utilizes the predicted residual to update the degradation model parameters iteratively. The simulation indicates that the proposed method achieves much better prediction accuracy and is more adaptable to the degradation characteristics of different prediction starting points of different battery individuals.