Prediction of Undrained Bearing Capacity of Skirted Foundation in Spatially Variable Soils Based on Convolutional Neural Network

Skirted foundations are widely used in offshore and subsea engineering. Previous studies have shown that soil undrained shear strength variability has a notable impact on probabilistic analyses of skirted foundation bearing capacity. This study proposes an efficient machine-learning method to predict the uniaxial capacity factors under pure horizontal moment loads, without relying traditional time-consuming random finite element methods. A two-dimensional convolutional neural network is adopted capture potential correlation between fields factors. The proposed CNN-based model exhibits satisfactory prediction performance with regard coefficients variation scale fluctuations two directions. Specifically, coefficient determination (R2) values exceed 0.97, while root mean square error (RMSE) remain below 0.13 for surrogate model. In addition, more than 96% predictions associated relative 5% or less, providing evidence 2D-CNN model’s performance.