Online learning ANN model for SoC estimation of the Lithium- Ion battery in case of small amount of data for practical applications

A battery's state of charge (SoC) estimation with accuracy is presented in this paper. Its calculation time is accelerated using online model approach (during operation) with optimal generalization. This novel idea minimizes the initial error of the SoC with the help of artificial neural network (ANN). It is realized by an online learning with offline update parameter estimating model for a battery equivalent circuit. Although degradation of Lithium-Ion (Li-Ion) battery cell is inevitable, accurate estimation of the actual state is needed to extend its life of usage. It is, however, difficult to calculate the parameters by conventional methods, because the parameters are nonlinearly involved in the mathematical notation of the battery impedance and all other variables related with SoC estimation.