Application of Artificial Neural Networks for Predicting Process Gas Flow Rate in the Direct Reduction Plant of Jahan Foolad Sirjan Company

In the direct reduction process of sponge iron production, access to and accurate prediction of various process gas parameters are of great importance. With the development of artificial intelligence technologies and their significant contributions to different areas of the steel industry, Jahan Foolad Sirjan Company has also adopted the use of this technology in the sponge iron production cycle. Accordingly, this study aims to estimate the process gas flow rate in the direct reduction process using artificial intelligence tools based on the physical conditions of the gas. First, process gas pressure and temperature data were extracted over a one-month period of plant operation. Then, using the artificial intelligence environment of MATLAB software, various artificial neural networks were trained with the available data, and the most accurate neural network was identified. The most important result shows that the developed artificial neural network is capable of estimating the process gas flow rate using gas temperature and pressure with an error of less than 2%.