Accurate state of charge estimation for lithium-ion battery using dual Uncsented Kalman filters

State of charge is a significant indicator with respect to the remaining capacity for the lithium-ion battery. Nonetheless, strong nonlinearity and time-varying attributes resulting from the complicated electrochemical reactions incur the tremendous difficulty to acquire the accurate state directly. To address the above problems, a novel estimation method based on unscented Kalman filter and dual-filters scheme is proposed. Concerning the nonlinearity features of lithium-ion batteries and the advantages of unscented transformation technique to deal with nonlinearity problems, the unscented Kalman filter approach is applied for states estimation and simultaneously in order to guarantee parameters variation of models adaptively with changes of batteries performance, the dual filters scheme is established to estimate the parameters and state of charge collaboratively. Through the validation by two dynamic driving cycles, it is shown that the state of charge not only could approach to the true values fast, but also the estimated errors are kept within the boundary of 2%.