The state of charge estimation for rechargeable batteries based on artificial neural network techniques

This paper presents an adaptive state of charge estimator for rechargeable batteries using the artificial neural network technique. That technique is based on that the charging current for any battery, in un-controlled current charging circuit, changes according to the battery state of charge (SOC). This proposed estimator will use the charging current, battery voltage samples and the time of each sample, from charging start, as an artificial neural network inputs and SOC as the output. The proposed estimator will be applied on Nickel-Cadmium battery model to test the validity of SOC neural network estimator to estimate the state of charge. Also, to know how the proposed estimator will be able to adapt with a new battery behavior such as capacity loss, the estimator will be tested in the case of a loss in capacity for the same Nickel-Cadmium battery model. The paper will depend on neural network and ANFIS using the simulations tools in MATLAB Program to make all required models, moreover, getting the training and testing data through a charging circuit model.