Tunnel Boring Machine AI

We are developing physics-based, data-driven and hybrid physics-data driven models for TBM-ground interaction. Recent physics-based model results for soft ground TBM torque and chamber pressure behavior can be found in the Publications section. We are also using machine learning approaches to develop data-driven models of TBM performance where the physics is simply too complicated. Using data from constructed tunnels throughout the world, our research has shown that data-driven models trained using machine learning are very successful at predicting ground conditions and identifying how to tunnel faster and with limited tunnel induced deformation. This is illustrated in the figure below.