Reliability analysis and inverse optimization method for floating wind turbines driven by dual meta-models combining transient-steady responses

Abstract

This study introduces an innovative framework aimed at enhancing the reliability, lifespan, and cost-effectiveness of floating wind turbines. By seamlessly integrating transient and steady-state response surface methods, this approach reduces the dimensionality of high-cost, high-fidelity simulation data for floating wind turbines. It effectively utilizes time-domain stochastic response history samples, addressing two critical aspects in reliability analysis, including statistical properties related to maximum loads relevant for static strength design and the distribution characteristics of vibration amplitudes pertinent to fatigue damage design. The processed data is used to construct a forward 1st metamodel, which is integrated into a multi-objective optimization algorithm. This algorithm determines optimal hyperparameters based on stochastic environmental conditions. Subsequently, an inverse 2nd metamodel is established, working in conjunction with the 1st metamodel, forming a dual structure for reliability analysis. Extensive Monte Carlo simulations and probability density analyses demonstrate the effectiveness of the adaptive flexible controller frequency strategy, obtained using the proposed framework, in reducing vibration-induced blade-tip-tower collision risks and long-term vibration damage, particularly at low to medium wind speeds. This significantly improves turbine stability and safety, enhancing overall performance. The proposed methodology provides valuable insights into the design, manufacturing, and operation of floating wind turbines.