Optimization of support system for urban deep excavations with machine learning techniques
Productivity in the construction sector, particularly in soil excavation, has stagnated for decades. The main advances were made 30 years ago with the development of finite element simulation tools. Today, planning an excavation for a building is still based on a long human process of trial and error. At the same time, important steps are being taken to unify data models for different construction professions. Important progresses are also being made in the field of soil cartography in Switzerland, such as SwissTopo. The development of these models will provide large amounts of numerical data describing the different parameters and stages of building construction and in particular excavation in a given soil environment. The underlying scientific question is about the usability of such data to improve methods and knowledge about geotechnical structure design. More specifically, the objective of the OptiSoil project is to use this data to apply artificial intelligence algorithms using machine learning techniques to develop innovative excavation design techniques that optimize stability, risk and cost constraints.