State of charge estimation of lithium-ion battery for electric vehicles using a neuro-fuzzy system

To accurately estimate the state of charge of a lithium-ion battery pack used in electric vehicles, a neurofuzzy system is proposed. The subtractive clustering is used to determine the structure and the initial parameters of the neuro-fuzzy system to reduce heuristic errors. The algorithm of adaptive neuro-fuzzy inference (ANFIS) is adopted to optimize the parameters of the neuro-fuzzy system. The training and validating data for the neuro-fuzzy system are sampled and calculated in the federal urban driving schedule (FUDS) test of the lithium-ion battery pack. The comparison with the model built by the subtractive clustering and least squares combined method illustrates that the neuro-fuzzy state of charge estimating model based on the subtractive clustering and ANFIS combined method has a higher accuracy.