Time series prediction in the case of nonlinear loads by using ADALINE and NAR neural networks

International Conference on Applied Sciences 2017

Ghiormez, L.^1 ; Panoiu, M.^1 ; Panoiu, C.^1 ; Tirian, O.^1

Politehnica University of Timisoara, Department of Electrical Engineering and Industrial Informatics, 5 Revolution Street, Hunedoara

331128, Romania^1

关键词: Electric arc furnace;    Electrical power supply;    Electrical supply;    Furnace transformers;    Measuring points;    Nonlinear load;    Three-phase loads;    Time series prediction;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/294/1/012026/pdf
DOI  :  10.1088/1757-899X/294/1/012026

来源: IOP

【 摘 要 】

This paper presents a study regarding the time series prediction in the case of an electric arc furnace. The considered furnace is a three phase load and it is used to melt scrap in order to obtain liquid steel. The furnace is powered by a three-phase electrical supply and therefore has three graphite electrodes. The furnace is a nonlinear load that can influence the equipment connected to the same electrical power supply network. The nonlinearity is given by the electric arc that appears at the furnace between the graphite electrode and the scrap. Because of the disturbances caused by the electric arc furnace during the elaboration process of steel it is very useful to predict the current of the electric arc and the voltage from the measuring point in the secondary side of the furnace transformer. In order to make the predictions were used ADALINE and NAR neural networks. To train the networks and to make the predictions were used data acquired from the real technological plant.