Supercapacitor thermal- and electrical-behaviour modelling using ANN
Abstract
The paper presents the development of a modelling tool for evaluation of the thermal and electrical behaviour of supercapacitors, using an artificial neural network (ANN). The principle consists of a black-box multiple-input single-output (MISO) model. The system inputs are temperature, current and supercapacitor values, and the output is the supercapacitor voltage. The relationship between inputs and output is established by the learning and the validation of the ANN model from experimental charge and discharge cycles of supercapacitors at different currents and different temperatures. Once the training parameters are known, the ANN simulator can predict different operational parameters of the supercapacitors. The update parameters of the ANN model are performed using the Levenberg-Marquardt method to minimise the error between the output of the system and the predicted output. This methodology using ANN networks may provide useful information on the transient behaviour of the supercapacitors taking into account thermal influences. Experimental results will also validate the simulation results.