On-line battery identification for electric driving range prediction

Hybrid and electric vehicles require accurate knowledge of the battery to make an educated guess about the expected electric driving range. Range prediction is complicated by the fact that batteries are subject to external influences and aging. Also the future driving behavior is often unknown. This paper presents a model-based identification method for online parameter estimation of Li-ion batteries. This allows range prediction to anticipate on all kinds of duty cycles as well as changing battery characteristics enforced by cell degradation. In the proposed methodology, parameters values of the battery model are frequently updated with the latest measurement data. Additional state observers are utilized to keep the dynamical states of the battery model up-to-date. Altogether, this offers robustness and accuracy under varying operating conditions and for the complete battery lifetime. Experimental results are presented to validate the proposed identification method.