Dynamic modelling of supercapacitor using artificial neural network technique

Because of the complex dynamic behavior of supercapacitor, its modeling must be based on parallel, distributed structures (each component has to represent a model of activity, distributed on many processing units), with learning capacity. For this purpose, the paper proposes a new feed forward artificial neural network structure with two hidden layers and with backpropagation training. The neural network provides, after activation, training, testing and reinitializing, output values with a total correlation of 0, 9426 compared with target values.