Model-Predictive Cubature Kalman Filter for Battery Core Temperature Estimation

Battery core temperature estimation plays a key role in developing efficient thermal management systems. This paper proposes a novel model-predictive cubature Kalman filter (CKF) to obtain high accuracy temperature estimation. Considering the thermal characteristics of cylindrical battery cells, a lumped two-layer thermal model is first formulated. The model parameters are identified offline based on a high-fidelity finite element model combined with particle swarm optimization. Then, the estimation law of model error is developed by the model prediction filter, which is integrated into the CKF to enhance the estimation accuracy. Simulation tests based on two automotive driving cycles are carried out with an initial temperature of 25 °C to verify the estimation performance using model-predictive CKF. And the model- predictive CKF has more accurate core temperature estimation compared to traditional CKF and adaptive CKF.