A surrogate dream optimization method for the back-analysis of mechanical parameters in wind turbine foundations

To address the challenge of rapid determination of mechanical parameters for wind turbine foundations under complex dynamic time-varying loads, a surrogate dream optimization method is proposed. This approach integrates the global-searching dream optimization algorithm (DOA) with a Gaussian process regression (GPR) model for surrogate numerical simulation, offering clear logic and easy implementation. In a back-analysis case of mechanical parameters of onshore wind turbine foundations, with 1000 numerical calculation calls, the method reduces the average settlement error by nearly 75% compared to the DOA method, demonstrating significantly higher accuracy. This provides a high-fidelity numerical model foundation for constructing reliable digital twins of wind turbine foundations, enabling more accurate safety evaluation and potential risk prediction on digital platforms, thereby significantly advancing the practical application of digital twin technology in intelligent manufacturing.