The application of an improved artificial neural network model for prediction of Cu and Au concentration in the porphyry copper-epithermal gold deposits, case study: Masjed Daghi, NW Iran

Modeling of geochemical data to predict elements is done with different methods. The proposed method in this research is the use of an intelligent model and pathfinder elements. In this study, drilling and sampling were done in two porphyry and epithermal mineralization of the Masjed Daghi porphyry copper deposit, and we used the data from the porphyry mineralization to predict copper and the data from the epithermal mineralization to predict gold. By using geochemical data and performing correlation and sensitivity analyses, copper and gold pathfinder elements (Pb, Zn, Ag, Mo, As) were determined. Then, using the data of pathfinder elements and an intelligent artificial neural network model, we predict the grade of gold and copper elements. The data of pathfinder elements were used as input and the grade of gold and copper elements were used as output of the model. In this research, the optimization of the artificial neural network is done using several optimization algorithms such as simulated annealing algorithm (SAA), firefly algorithm (FA), invasive weed optimization algorithm (IWO) and shuffled frog leaping algorithm (SFLA). Comparing the results showed that ANN-SAA (Combining ANN with SAA) performs better than other built models. This superiority was evident both in the porphyry and epithermal mineralization. R2 and MSE of ANN-SAA model for Cu prediction were 0.8275 and 0.0303 for training data, 0.7357 and 0.0371 for testing data respectively. Also, R2 and MSE of ANN-SAA model for Au prediction were 0.6713 and 0.0463 for training data, 0.7040 and 0.0333 for testing data respectively.

The application of an improved artificial neural network model for prediction of Cu and Au concentration in the porphyry copper-epithermal gold deposits, case study: Masjed Daghi, NW Iran Keywords:

Prediction of Cu and Au , Artificial Neural Network , Evolutionary algorithms , Arasbaran metallurgical zone , Porphyry copper deposits