Physics-informed Data-driven Modelling in Geotechnical Engineering

Abstract:
Physics-informed data-driven modelling has been extensively used across various domains, but its computational efficiency and practicability in engineering practice often incur scepticism. Particularly, its application in geotechnics still remains in infancy. To answer these questions, we enhance current physics-informed data-driven modelling for forward and inverse analysis of geotechnical problems by integrating adaptive sampling, domain decomposition, Ritz method, transfer learning and customised optimizer and loss functions. Its feasibility is demonstrated by applying it to 1D and 2D consolidation equations, and footing problems. The results indicate that enhanced physics-informed data-driven modelling enables solving the time-dependent, elastic and elasto-plastic problems. This framework allows for flexible data assimilation, including sparse in-situ monitoring and historical data, to derive accurate solutions for problems at hand. This capability offers a promising and cost-effective strategy for sensor installation, real-time prediction, model parameters identification and digital twinning in engineering practice.