Recurrent Neural Network based Data-Driven SOC Estimation in Lithium-Ion Battery

Lithium-Ion batteries are used popularly in many technologies used in day-to-day life. Engineering research on Lithium-Ion battery efficiency assessment is underway due to its large size acceptance. This paper develops the SOC estimation using the data driven method. Recurrent Neural Network based implementation is developed using the dataset val-ues obtained from a experimental data of 3Ah bat-tery LG HG2. Parameters including voltage, current, temperature, average voltage and average current is observed in the experiment available for research. A dataset that includes different battery parameters with temperature, voltage and current are mapped to the State of Charge of the battery. This dataset is used to train the Recurrent Neural Network in this paper for SOC estimation. The major goal of this research is to establish the appropriate Recurrent Neural Net-work (RNN) design procedure for a SOC estimator. Performance evaluation of the algorithm this used is done and tabulated. Python scikit learn toolbox is used to develop the algorithm and results is found to be satisfactory.