Applied Machine Learning for Geometallurgical Throughput Prediction—A Case Study Using Production Data at the Tropicana Gold Mining Complex

Abstract

With the increased use of digital technologies in the mining industry, the amount of centrally stored production data is continuously growing. However, datasets in mines and processing plants are not fully utilized to build links between extracted materials and metallurgical plant performances. This article shows a case study at the Tropicana Gold mining complex that utilizes penetration rates from blasthole drilling and measurements of the comminution circuit to construct a data-driven, geometallurgical throughput prediction model of the ball mill. Several improvements over a previous publication are shown. First, the recorded power draw, feed particle and product particle size are newly considered. Second, a machine learning model in the form of a neural network is used and compared to a linear model. The article also shows that hardness proportions perform 6.3% better than averages of penetration rates for throughput prediction, underlining the importance of compositional approaches for non-additive geometallurgical variables. When adding ball mill power and product particle size, the prediction error (RMSE) decreases by another 10.6%. This result can only be achieved with the neural network, whereas the linear regression shows improvements of 4.2%. Finally, it is discussed how the throughput prediction model can be integrated into production scheduling.

Prediction of energy consumption in grinding using artificial neural networks to improve the distribution of fragmentation size [Predicción del consumo de energía en la molienda utilizando redes neuronales artificiales para mejorar la distribución del tamaño de la fragmentación]

“…Furthermore, a learning rate of 0.01 and 1500 epochs were considered within the hyperparameters. Both and Dimitrakopoulos (2021), in their research to build a geometallurgical performance prediction model based on ball mill data, considered the recorded energy consumption, feed particles and product particle size, for which They used a neural network, comparing it with a linear model. In their study they used a single hidden layer neural network structure comprising 30 neurons offering the most stable predictions and showing a performance prediction error variation of 10.6%.…”

Recursive Feature Elimination and Neural Networks Applied to the Forecast of Mass and Metallurgical Recoveries in A Brazilian Phosphate Mine

“…Both and Dimitrakopoulos [30] applied neural networks to forecast throughput and compared the results against the throughput predicted by linear regression. The NN showed better results and the predictions obtained with its use can be integrated into production scheduling; • Gholami et al [31], used neural networks to forecast four geometallurgical variables in a copper mine, using as inputs particle size distribution, collector and frother concentrations, solids content, pH and mineralogical variables; • Jorjani et al [32] created a neural network model to forecast La, Ce, Y and Nd recoveries from an apatite concentrate.…”