Artificial Neural Networks for Prediction of Hot Metal Production in a Blast Furnace

A blast furnace is a chemical reactor used in the steel industry to produce molten iron or hot metal. The size of this reactor is variable and can be more than 30 m high. It is coated with metal on the outside and refractory material on the inside. The reactor operates at high temperature and pressure. It is fed with coke, iron ore and fluxes in the upper part and air and auxiliary fuels are injected in the lower part. The rising gases react with the descending solids and melt the material. The production of pig iron also produces slag, which is normally used to make cement. This scientific article reports on the successful application of artificial neural networks in pig iron production. The neural network was modelled in MATLAB using 23 operational variables with 100 neurons. The validation of the mathematical model was carried out through statistical tests in the MINITAB software, which ensure the necessary statistical certainty for the validation of its application on an industrial scale.