The past decade has shown a rapid increase in the successful application of Machine Learning techniques for a variety of challenging tasks. Potential for this is also seen in the automatic rockmass behavior classification of tunnel boring machine (TBM) advance-data. This study compares the performance of two kinds of Artificial Neural Networks (ANN) - a Multilayer Perceptron (MLP) vs. a Long-Short-Term Memory (LSTM) Network - for this task. The data originates from the exploratory tunnel Ahrental – Pfons of the Brenner Base Tunnel which is currently under construction. The goal of gathering as much data as possible from the encountered geology is to transfer this knowledge from the exploratory tunnel to the main tunnel tubes. Results show that both ANNs are capable of classifying rockmass behavior only based on TBM advance data, however, the LSTM outperforms the MLP in several of the test-data samples.
|Title of host publication||Rock Mechanics for Natural Resources and Infrastructure Development|
|Subtitle of host publication||Full Papers: Proceedings of the 14th International Congress on Rock Mechanics and Rock Engineering (ISRM 2019)|
|Editors||Sergio A.B. da Fontoura, Ricardo Jose Rocca, José Pavón Mendoza|
|Publication status||Published - 2019|
Erharter, G. H., Marcher, T., & Reinhold, C. (2019). Comparison of artificial neural networks for TBM data classification. In S. A. B. da Fontoura, R. J. Rocca, & J. Pavón Mendoza (Eds.), Rock Mechanics for Natural Resources and Infrastructure Development: Full Papers: Proceedings of the 14th International Congress on Rock Mechanics and Rock Engineering (ISRM 2019) CRC Press.