Prediction of Forward Tunnel Face Score of Rock Mass Classification for Stability by Applying Machine Learning to Drilling Data

Abstract

In tunnel construction, it is important to ensure economic efficiency and safety when collecting data on ground condition variations during construction. Therefore, in recent years an increasing number of exploration methods using rock drills in front of tunnel faces have been developed. In this paper, a new exploration method called TFS-learning was proposed; this method uses machine data of a rock drill. Further, TFS identifies the relationship between the drilling machine data during the perforation of the blast hole and the assessment score of the tunnel face through machine learning to predict the distribution of assessment scores on the tunnel face. In this study, TFS-learning is focused on drilling data of the applied A tunnel. Based on the analysis result, a regression equation is derived, from which, the accuracy of the predicted result of the forward face assessment scores is verified.

Therefore, the face assessment scores could be roughly predicted in the section with similar rock types by machine learning; however, it was also found that prediction accuracy decreases when the rock type changes.