Unplanned dilution prediction in open stope mining: developing new design charts using Artificial Neural Network classifier

Minimizing dilution is essential in open stope mine design as excessive unplanned can compromise the operation's profitability. One of main challenges associated with empirical graph method used to stopes how determine boundary zones objectively. Hence, this paper explores implementation machine learning classifiers bridge gap conventional method. Stope performance data consisting (unplanned dilution), modified stability number, and hydraulic radius were compiled from a located Kazakhstan. First, methods assess dilution. Next, Feed-Forward Neural Network (FFNN) classifier was implemented predict each level Overall, FFNN results indicated that 97% surfaces correctly classified, indicating an excellent classification performance, while did not show good performance. In addition, outputs plot new graphs probabilistic interpretation illustrating its practicability. It concluded FFNN-based could be useful tool for underground mines.