.png){kind=icon}
.png){kind=icon}
Return
Physics-informed data-driven modelling and computation in geotechnics
.png){kind=icon}
PRE
AI
DOI：10.1080/17499518.2026.2641683.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Physics-informed data-driven (PIDD) modelling has attracted considerable attention across a wide range of disciplines. Our recent studies indicate that PIDD modelling provides a concise and elegant means of capturing soil behaviour directly from data. However, its feasibility in representing more complex soil responses requires further validation. PIDD computation also offers a promising alternative to the finite element method (FEM), though its capability in addressing multi-phase coupling problems remains to be enhanced. Moreover, its potential integration with field data – particularly for data assimilation and inverse analysis – holds significant promise. This paper provides a detailed discussion of PIDD modelling and computation in geotechnics. A novel thermodynamically consistent hierarchical learning framework is introduced for the automatic identification of internal variables and prediction of stress–strain responses in granular soils, followed by its integration with FEM to validate its applicability to boundary value problems. The discussion further extends to the development of a PIDD-based solver for canonical geotechnical problems, including one- and two-dimensional consolidation and footing analyses, as well as its application to inverse analysis. The paper concludes with a comprehensive summary of PIDD in geotechnics, offering insights into future directions for advancing this research area within the geotechnical community.
Keywords:
Physics-informed
data-driven
neural networks
constitutive modelling
computational geotechnics
Journal
G
IF:
0
Papers:
36
Citations:
0