Shear Wave Velocity Prediction Using Machine Learning in Brazilian Marine Clay

This study evaluates the feasibility of using machine learning (ML) algorithms to predict shear wave velocity (Vs) in marine clay based on geotechnical parameters. In situ measurements of Vs were collected from Santos Basin, Brazilian coast using the a remotely operated seabed drilling system equipped with in situ seismic testing tools to train predictive ML models. The dataset wa processed to remove errors and ensure representativeness, and model validation was performed through cross-site predictions, including blind testing at an untrained location. The results demonstrate that ML algorithms can achieve higher accuracy and consistency compared to traditional empirical correlations, especially in capturing key geotechnical trends. The analysis also highlights the influence of specific parameters on Vs prediction, providing insights for optimizing site investigation strategies. Furthermore, the applicability of the models across diverse offshore conditions reinforces their robustness. This research contributes to the advancement of AI-driven approaches in offshore geotechnical engineering, supporting more efficient site characterization and reducing uncertainties in practice.