Rock collapse forecasting: A novel approach based on the classification of micro-acoustic signals in the wavelet domain

This paper proposes a novel approach for the rock collapse forecasting based on the automatic classification of micro-acoustic emissions in the wavelet domain. Solutions present in the literature are surpassed in two main directions. First, we designed a novel and comprehensive set of features extracted from micro-acoustic emissions based on the Discrete Wavelet Transform. Second, we consider and contrast several machine learning classification techniques. We evaluated the accuracy of the proposed approach on real-world data acquired by a real-time monitoring system for rock-collapse forecasting deployed in Northern Italy. Experimental results demonstrate the effectiveness of what proposed.