Comparative analysis of ai models in the modeling of flotation process

Abstract

This paper presents a comparative overview the modeling results of flotation process from the Veliki Krivelj plant. Within the research, a total of ten models were formed using the various methods of soft computing-fuzzy logic, artificial neural networks and hybrid system ANFIS. It was found that the flotation concentration process is best modeled using artificial neural networks, where the best correlation coefficients between the actual and predicted values of copper content in concentrate and tailings, as well as the copper recovery in concentrate are achieved. They also gave the minimal RMSE in all cases.