A variable multi-time-scale based dual estimation framework for state-of-energy and maximum available energy of lithium-ion battery

To lower the computation burden and enhance co-estimation reliability under unpredicted operating conditions, this paper presents a novel variable multi-time-scale based dual estimation framework for state-of-energy (SOE) and maximum available energy. Through forgetting factor recursive least squares (FFRLS) based model parameters identification method, the first-order RC model is online built firstly to simulate battery dynamics. Subsequently, identified model parameters are inputted into an adaptive extended Kalman filter to predict SOE. Meanwhile, with battery data and two estimated SOE, inaccurate maximum available energy can be further updated by FFRLS when energy accumulation reaches pre-defined threshold. Especially, to determine the optimal macro time-scale considering co-estimation performance comprehensively, a multi-objective decision analysis method by fusion of analytic hierarchy process and the entropy weight is innovatively proposed. The dual estimation accuracy and robustness ability of the proposed framework are verified with experimental data of Federal Urban Driving Schedule tests conducted under various temperatures, whose results show that the presented method has satisfactory co-estimation accuracy and robustness ability. Furthermore, the comparison with other algorithms not only indicates the necessity of maximum available energy updating on SOE prediction but also the superiority of the presented framework on dual estimation accuracy and computational cost.