Artificial Intelligence Model for Prediction of Local and Main FALL in caving Panel of Bord and Pillar Method of Mining

Abstract

Depillaring with caving method of mining is a common practice in Indian coalfields and so is the occurrence of fall in goaf area, which can be considered as a boon in disguise as it allows wining of coal from large reserves but this becomes a curse just because of its unpredicted occurrence. Various empirical and statistical models are developed after idealization of several complicated mechanisms but they are not able to predict roof fall accurately especially in caving panels. Therefore, a new approach based on Artificial Intelligence is used to predict the sequence of local and main fall in caving panel taking into account a host of geotechnical and mining parameters of the mine. Mathematical equations and hidden calculations of artificial neural networks are known to have the capability of learning and analyzing records endlessly. Two different models have been deployed after optimal hyper parameter optimization to predict the occurrence of fall and to characterize the nature of fall (local or main) with considerable and reliable accuracy.