Predicting mine dam levels and energy consumption using artificial intelligence methods

Four machine learning algorithms (artificial neural networks, a naive Bayes' classifier, a support vector machines and decision trees) were applied for a single pump station mine to monitor and predict the dam levels and energy consumption. This work was undertaken to investigate the feasibility of using artificial intelligence in certain aspects of the mining industry. If successful, artificial intelligence systems could lead to improved safety and reduced electrical energy consumption. The results show neural networks to be more efficient when compared with support vector machines, a naive Bayes' classifier and in particular, decision trees in terms of predicting underground dam levels. Artificial neural networks showed 60% accuracy, out-performing support vector machine, naive Bayes' classifier and decision trees. For the prediction of water pump energy consumption, an artificial neural network and a naive Bayes' classifier had the same accuracy of 99.0%, whereas a support vector machine and decision trees achieved a lower accuracy.