The anxiety of driving range for drivers has become one of important problems to further progress. The imprecise evaluation of battery state is regarded as a crucial cause for restricting the accurate estimation of driving range, especially for state of energy (SOE). The inaccurate SOE can be attributed to uncertain driving condition and imprecise battery model. In this article, a coupling model-data driven method is proposed for improving the precision and adaptiveness for SOE estimation, where an equivalent-circuit-model-based method is applied for predict SOE under given condition, and a well-trained Elman network is used for adaptively correcting the predicted results. The proposed method is validated with high accuracy, within 3%, and robustness compared against sole model-based method. Further feasible promotion for online estimation of driving range is also discussed.