Electric current prediction for the nonlinear high power loads using Narx neural networks

This paper presents a study on the possibility of making a prediction for the electric current of nonlinear high power loads. One of the most significant nonlinear loads is the electric arc furnace (EAF). EAF is widely used in steel production and introduces many disturbances in the electric power supply. The prediction of the electric current provides information about the process that is useful in process control. For making the prediction recurrent Narx neural networks were used. The prediction was made using Matlab Neural Network Toolbox. To train the artificial neural network and to perform the prediction, measurements of currents and voltages were used. The measurements were made from the secondary transformer that supplies an EAF.