THE USE OF ARTIFICIAL NEURAL NETWORKS IN THE CONTROL OF ELECTRIC ARC FURNACES

Abstract

Today, control systems have become an important branch of science in parallel with the increase of production and quality needs. There are purpose-specific automatic control systems and algorithms controlling them for production in industrial facilities.
 In this study, modeling electric arc furnace scrap melting plant, which has an essential place in the iron-steel industry has been made using artificial neural networks. The facility where the study is carried out is in active production and controlled by classical algorithms. Artificial neural networks were trained using the data taken over the current control system and pressure sensors attached to the electrodes and the modeling and control of the arc furnace with the trained network was carried out. 
 The software developed with an artificial neural network to control the electrodes used in electric arc furnaces provided 98% success in monitoring the system including the operator’s intervention out of the algorithm. All input and output data of an active production facility were copied to the network with the developed software. Since this software does not require various calculations, calibrations and parameter changes, it responds faster than the classical control algorithm used in the factory.