In the quest to harness the power of offshore wind, engineers face a formidable challenge: the unpredictable nature of seabed soils. A recent study led by Hong Xia of the Zhejiang College of Construction, published in *Advances in Civil Engineering* (translated as *Advances in Civil Engineering*), offers a promising solution to this age-old problem. By leveraging the power of machine learning, Xia and his team have developed a novel method for predicting the undrained shear strength (su) of seabed soils, a critical parameter for the design of offshore wind turbine foundations.
The study addresses a longstanding issue in the field: the limited availability of labeled data for piezocone penetration tests (CPTU), which are commonly used to estimate su. “The complexity of seabed soils and the lack of labeled data have made it difficult to develop accurate predictive models,” Xia explains. To overcome this challenge, the researchers employed a self-training semi-supervised learning (SSL) strategy, which allows the model to learn from both labeled and unlabeled data.
The results are impressive. The proposed method reduces the mean absolute error (MAE) from 10.92 to 9.70 kPa and the root mean square error (RMSE) from 17.81 to 15.69 kPa compared to baseline models. Moreover, the incorporation of uncertainty quantification enhances the stability and reliability of the predictions. “This method not only improves the accuracy of our predictions but also provides a measure of confidence in those predictions,” Xia says.
The study also sheds light on the relative importance of different input parameters in predicting su. Using SHapley Additive exPlanations (SHAP)-based interpretability analysis, the researchers found that tip resistance (qc) is the most influential input. This finding could guide future data collection efforts and model development.
The implications of this research are significant for the energy sector, particularly for the offshore wind industry. Accurate prediction of su is crucial for the design and optimization of offshore wind turbine foundations, which in turn affects the overall cost and feasibility of offshore wind projects. As the world increasingly turns to renewable energy sources, advances in offshore wind technology will be essential to meeting global energy demands.
Looking ahead, the researchers hope to see their method adopted by industry practitioners and further refined through collaboration and data sharing. “We believe that this method has the potential to revolutionize the way we approach seabed soil analysis and offshore foundation design,” Xia says. With continued research and development, the dream of harnessing the full power of offshore wind may soon become a reality.