Rock recognition and identification for selective mechanical mining: a self-adaptive artificial neural network approach

Abstract In situ characterisation of rock is crucial for mine planning and design. Recent developments in machine learning (ML) have enabled the whole learning, reasoning, decision-making process to be more efficient accurate. Despite these developments, application ML rock-cutting at an early stage due lack mining applications mechanised excavation leading limited availability data sets expert knowledge required when fine-tuning models. This study presents a novel approach identification during mechanical by applying self-adaptive artificial neural network (ANN) model classify types selective cutting, which datasets from two cutting operations (actuated disc (ADC) oscillating (ODC)) were employed test train model. The was also configured with Bayesian optimization algorithm determine optimal hyperparameters automated manner. By comparing performance each evaluation, trained identify best set hypermeters uncertainty minimal. Further testing indicated very accurate classifying ADC as accuracy, recall, precision all equal unity. Some misclassifications occurred ODC ranging 0.68 0.99. promising results proved robust scalable tool enabling interpretation performed precisely, selectively, efficiently. Since requires significant energy, any improvement matching characteristics mass will increase productivity, energy efficiency reduce cost.