A Case Study of Prediction and Analysis of Unplanned Dilution in an Underground Stoping Mine using Artificial Neural Network

Abstract

Stoping method has been acknowledged as one of the typical metalliferous underground mining methods. Notwithstanding with the popularity of the method, the majority of stoping mines are suffering from excessive unplanned dilution which often becomes as the main cause of mine closure. Thus a reliable unplanned dilution management system is imperatively needed. In this study, reliable unplanned dilution prediction system is introduced by adopting artificial neural network (ANN) based on data investigated from one underground stoping mine in Western Australia. In addition, contributions of input parameters were analysed by connection weight algorithm (CWA). To validate the reliability of the proposed ANN, correlation coefficient (R) was calculated in the training and test stage which shown relatively high correlation of 0.9641 in training and 0.7933 in test stage. As results of CWA application, BHL (Length of blast hole) and SFJ (Safety factor of Joint orientation) show comparatively high contribution of 18.78% and 19.77% which imply that these are somewhat critical influential parameter of unplanned dilution.