Separator equipment performance of iron ore and coal using experimental and ANN-based analysis

The present study deals with the development of a new separator for the separation of iron ore and coal of a size fraction of -4 + 0 mm individually. Particles of iron ore with size fraction - 2 + 0 mm and finer coal were separated separately using screen mesh with an aperture size of 2 mm. The operating characteristics of the screen's upward slope and the screen's vibration frequency of the new separation equipment can be easily modified. In this study, moist iron ore and coal segregation have been carried out for various separation angles and frequencies, and test results of moist iron ore and coal were compared based on their moisture content and density. Also, for the prediction of results, artificial neural network (ANN) modeling and regression analysis were implemented. The R-square value for regression analysis of experimental results was found higher than 85.60% and 88.50% for coal and iron ore respectively. The R-square value for the ANN mathematical model of experimental results was found higher than 99.10% and 98.24% for coal and iron ore respectively. The comparison of the regression model, the ANN model's mathematical modeling results, and the test results for the separation of moist iron ore and coal hold a strong correlation. For validation, residual analysis was also performed on the separation of moist iron ore and coal regression and ANN models. Including improved accuracy, reduced computational time, and enhanced predictive capabilities. The residual probability plot's results for homoscedasticity, low standard deviation, normality, and independence demonstrate that, under all experimental settings for iron ore and coal, the developed artificial neural network (ANN) model outperforms the regression model in terms of prediction accuracy.