Simultaneous State and Parameter Estimation of Li-Ion Battery with One State Hysteresis Model Using Augmented Unscented Kalman Filter

An online technique for simultaneous state and parameter estimation of an electric or hybrid vehicle battery using an augmented unscented Kalman filter (AUKF) is proposed. Battery state of charge (SOC), the prime state variable, is estimated. In addition the battery internal resistance, time constants, and resistance of the resistance-capacitance (RC) pairs are also estimated. While the two RC pairs of the equivalent circuit capture voltage dynamics, the hysteresis effect will capture the difference in the polarization of the electrodes between charge and discharge. A separate state representing the dynamics of the hysteresis voltage is included in the nonlinear state-space representation of the lithium ion battery model. The parameters are judiciously chosen to keep the overall estimation technique robust enough. It is demonstrated that the inclusion of parameters in a state-space representation of a battery model can generate better SOC estimation accuracy and the claim is corroborated with correlation between experimental and simulation results.