An Adaptive Neuro-Fuzzy Inference System for Predicting Survivability Rate in Underground Mining Accident

Abstract

Underground mining has been characterized by high accidental rates over the years. Many efforts have been put in place to decrease the response time in evacuation situations after accidents. These efforts are reactive mechanisms rather than proactive mechanisms. Reactive mechanisms eventually end up losing valuable lives and properties. There is a need to develop a proactive mechanism to reduce the impact of underground mining accidents. This paper developed an Adaptive Neuro-Fuzzy Inference System (ANFIS) to predict the survivability rate of fall of ground in underground mining site using AngloGold Ashanti as a case study. Fuzzy rules were formulated by training Artificial Neural Network using such parameters like working shifts, types of rocks, nature of the rock, rock roof thickness and nature of stope to predict the rate of survivability as either high or low in accidental situations. These parameters were normalized by assigning them values between 0 and 1 to train the ANFIS model. The model’s predictions were compared with some recorded data for verification which proved to be 90% accurate. The implemented model will help policy-makers to plan for an inevitable accident to reduce the impact of the accident.