Intelligent Modeling of Cement Plant Mill Unit Using Artificial Neural Networks and Real Data

Abstract

The use of multi-compartment tube mills in closed circuit with an air separator is prevalent for grinding clinker, pozzolan and gypsum mixtures with certain percentages in cement factories. To produce cement in Arta Ardabil Cement Factory in Iran, this circuit is also used. The final product of this factory will be Portland Pozzolana Cement (PPC). In this paper, a method for modeling the desired circuit will be proposed and presented by Multi-layer Feed-Forward Neural Networks (ML-FF-NN). For modeling, many examples of the routine working process of the milling circuit on the production line have been considered. After data preprocessing operation, the model was designed and simulated using a feed-forward neural network under the backpropagation training algorithm. The results of training, testing and validation of the obtained model are satisfactory. Also, comparing the obtained model results with real data shows its high-performance accuracy and reliability.