Artificial neural networks to prediction fuel rate in the blast furnace operation.

Abstract

This paper proposes the use of artificial neural networks for the prediction of fuel consumption in the blast furnace. For this purpose, a dataset of 270 records, with 19 input variables were considered, based on the historical data of operation from the years 2014 to 2017 of a blast furnace of a Brazilian steel mill, and it was verified that model presented good results with correlation coefficient of 0.837, consisting of an input layer with 19 neurons, intermediate layer with 19 neurons and output layer with 1 neuron.