Current Practice and Tendency of Machine Learning for Ground Movement Predictions and Buildings Impact Assessment induced by Shield Tunnelling

With the increasing urbanization and the demand for efficient transportation infrastructure, shield tunnelling has become a prevalent method for constructing underground transportation networks. In many cases, several existing buildings are adjacent to the underground construction areas. Inevitably, those buildings may suffer a potential impact due to the construction, particularly tunnel construction associated with
excessive ground movements. Therefore, this article aims to demonstrate the methodology and approach in prediction of ground movements induced by shield tunnelling in Bangkok subsoil conditions and risk assessment carried out for damage categories of the existing buildings in practice. Following the three stages of the assessment procedure, the results from this study indicate that using surface ground movement for building damage assessment in stage 1 and stage 2 does not yield a conservative prediction in some cases. Therefore, buildings located within the pile-tunnel interaction zone should be further investigated in the third stage. Additionally, this paper aims to explore applications of machine learning techniques in the field of building impact assessment induced by shieldtunnelling.