Prediction of Slope Stability Coal Mine using Adaptive Neuro-Fuzzy Inference System (ANFIS)

ABSTRACT: Slope stability is a critical aspect in mine planning, particularly in coal mining operations in Indonesia. In many cases, coal mine that have instability events despite in the initial analysis indicating stable conditions. This often occurs due to uncertainty factors. This study aims to predict slope stability in coal mines using the Adaptive Neuro-Fuzzy Inference System (ANFIS), focusing on whether the slope is stable or unstable. The input parameters considered include unit weight, cohesion, friction angle, overall slope height, and slope angle. The data is divided into training and testing sets in an 80:20 ratio. Two models are developed: ANFIS-Model 1 and ANFIS-Model 2. ANFIS-Model 1 first predicts the factor of safety (FOS) and probability of failure (POF) and then determines the slope's stability. On the other hand, ANFIS-Model 2 directly predicts the slope's stability. Both models use a fuzzy inference system with subtractive clustering. The results show that ANFIS-Model 1 achieves an accuracy of 65%, precision of 50%, recall of 86%, and F-1 score of 63%. ANFIS-Model 2 outperforms ANFIS-Model 1, with an accuracy of 80%, precision of 92%, recall of 79%, and F-1 score of 85%. A chart representing the results of ANFIS-Model 2 is also presented.