ABSTRACT
Recent studies showed that the artificial neural network (Shahin et al. 2004, Jaksa et al. 2008) can be employed successfully for the prediction of settlements induced by tunnelling (Neaupane & Adhikari 2006, Yoo & Kim 2007). Compared to the usual design methods, they allow analyzing a larger quantity of data by integrating more complicated nonlinear combinations (Kim et al. 2001). The excavation parameters, particularly those of TBM operation, are more easily analyzed with ANN (Javadis 2006, Santos & Celestino 2008). This approach is less severe on the consideration of the geological conditions for settlements prediction; it does not require an identification of the soils’ constitutive laws and parameters to represent the variability of the ground (Yoo et al. 2009).
