Session: 10-03-01: Piles, Spudcans and Others

Submission Number: 157032

Transfer Learning Enhanced Composite Physics-Informed Neural Network：Application to Offshore Geotechnical Experiments 

The uniaxial compression test is a critical method frequently employed to assess the mechanical properties and deformation behavior of seabed soils. It plays a pivotal role in understanding soil strength characteristics, compression indices, and failure mechanisms, which are essential for the design and stability analysis of offshore structures. When conducting a large number of experiments on seabed soils from different regions, numerical simulation emerges as an efficient alternative. In recent years, advancements in deep learning have demonstrated the significant potential of Physics-Informed Neural Networks (PINNs) for simulating this process. In offshore geotechnics, the governing equations centered on the solid momentum equation and constitutive equation are highly complex. The Composite Physics-Informed Neural Network (CPINN) overcomes the pathological gradient issues inherent in traditional PINNs architectures. This study proposes a Transfer Learning Enhanced Composite Physics-Informed Neural Network (TL-CPINN) approach. This method exclusively applies transfer learning to the general solution network within a pre-trained CPINN, while keeping the distance network and particular solution network unchanged. Without requiring any additional datasets, this approach effectively inherits the boundary information and physical knowledge learned by the prior CPINN. Using the numerical simulation of the uniaxial compression test of elastic soils as an example, TL-CPINN achieves rapid simulation of mechanical behavior for different soil materials after only a single full CPINN training session. This research highlights the unique memory capabilities of PINNs compared to other numerical simulation methods, significantly enhancing the speed of numerical simulations.

Presenting Author: Yong Fu Southern University of Science and Technology

Presenting Author Biography: Dr Yong fu obtained his PhD degree in the National University of Singapore in 2018, andworked as a research fellow during 2018 and 2020 before he joined the Department of OceanScience and Engineering, Southern University of Science and Technology, Shenzhen, China, asan Assistant Professor, His research interest and expertise include offshore geotechnicalengineering (ie. torpedo anchors, spudcans and pipelines), underground space technology(i.e. development of horizontal DCM equipment), physical experiments (i.e. laboratory.centrifuge and field experiments), large deformation finite element modelling (i.e. CEl andRIT5S methods). In recent years, Dr fu has published more than twenty high-qualityinternational journal papers in world-renowned geotechnical journals such as Geotechnique.Canadian Geotechnical Journal, Computers and Geotechnics, Geotechnical Testing Journal. lnaddition, he also served as the reviewer of several international journals.