Artificial neural networks in streamlining serviceability assessment of single shell TBM tunneling

Artificial neural networks (=ANN) is the upcoming technology of the recent years. It improves the performance of search in internet, enables autonomous cars and enhances interaction with electronic devices. However, the reliability of ANN’s heavily depends on the data base neural networks are trained on. In tunneling there are rare applications of neural networks although they can mitigate labor intensive, tedious and error-prone information handling for e.g. as-built documentation and geotechnical measurement. Tunneling constitutes a harsh environment for gathering the indispensable representative data. The contribution shows the necessity of constituting a data set with sufficient quantity and representative quality for the training of networks. It is a decisive prerequisite for successful application of machine learning which can be applied for the assessment of serviceability of single-layer lining. Additionally, its role in crack detection in geotechnical tests is emphasized. The contribution points out different ANNs ready to be readapted for tunneling purposes. The data for setting up the data base needed for network training is gathered in Austrian TBM projects using 3D laser scanners and image acquisition systems during geotechnical testing.