Random Forest-Based Prediction Model for Stiffness Degradation of Offshore Wind Farm Submarine Soil

Offshore wind power is a hot spot in the field of new energy, with foundation construction costs representing approximately 30% of the total investment in wind farm construction. Offshore wind turbines are subjected to long-term cyclic loads, and seabed materials are prone to causing stiffness degradation. The accurate disclosure of the mechanical properties of marine soil is critical to the safety and stability of the foundation structure of offshore wind turbines. The stiffness degradation laws of mucky clay and silt clay from offshore wind turbines were firstly investigated in the study. Experiments found that the variations in the elastic modulus presented “L-type” attenuation under small cyclic loads, and the degradation coefficient fleetingly decayed to the strength progressive line under large cyclic loads. Based on the experimental results, a random forest prediction model for the elastic modulus of the submarine soil was established, which had high prediction accuracy. The influence of testing the loading parameters of the submarine soil on the prediction results was greater than that of the soil’s physical property parameters. In criticality, the CSR had the greatest impact on the prediction results. This study provides a more efficient method for the stiffness degradation assessment of submarine soil materials in offshore wind farms.