The present Thesis, valid as dissertation for the Master in High Performance Computing, summarizes the main results obtained during the course of the final project of the Master. The project was carried out as a collaboration between SISSA MathLab and the Lombardi Group, an italo-swiss company of civil engineer, and consists in a preliminary study about the development of numerical methods for supporting decisions regarding the safety of tunnels in Italy’s highway system. Two are the main points of interest considered:
- Application of algorithms such as Dynamic Mode Decomposition (DMD) and its variants to predict the time evolution trend of defects in a tunnel;
- Definition of an evaluation metric for the risk of a tunnel (Class of Attention) and implementation of Neural Networks to automate and speed up its computation.
In addition, ideas to reduce the problems found in the available data have been proposed, as well as best practice recommendations for future data collection and storage.
Thesis not available.