ANALYSIS OF THE POSSIBILITY OF INCREASING THE EFFICIENCY OF FILTRATION PROCESS OF FINELY GROUND IRON ORE MAGNETITE CONCENTRATE

Abstract

The article investigates the possibility of increasing the efficiency of the filtration process of finely ground iron ore magnetite concentrate in order to ensure stable moisture indices of the raw material required for the production of highquality DR pellets. The main objective of the study is to improve the efficiency of the filtration process by developing a mathematical model based on the gradient boosting method to predict the residual moisture of the concentrate and to optimize the technological parameters of the operation of KDF‑90 ceramic disc vacuum filters. Experimental studies were carried out on industrial KDF‑90 filters at PJSC «CGZK» using a dataset of 199 industrial samples containing 20 features, including basic and combined technological parameters. To build the model, a Gradient Boosting Regressor was used with feature standardization, a stratified 80/20 train–test split, and hyperparameter optimization according to the criterion of maximizing the coefficient of determination R² on the test set. The optimized model provided R² = 0.4091 on the test data with a mean absolute error MAE = 0.3380%, which outperforms linear models and is comparable with other ensemble methods, while the root mean square error was RMSE = 0.4078%. It was established that the strongest impact on residual moisture is exerted by the content of the -0.056 mm size fraction, the vacuum in the drying zone, and the combined parameter given by the product of total iron content and the -0.040 mm fraction. The practical significance of the developed model lies in the possibility of reducing fluctuations in concentrate moisture, increasing the stability of the pelletizing process, and providing recommendations for the implementation of an automated system for moisture prediction and control under industrial conditions.