10 MW급 부유식 해상풍력발전기용 FCS의 구조설계 민감도 해석 및 메타모델 비교연구
Meta-modeling and design sensitivity analysis of detachable mooring system for 10 MW-class floating offshore wind turbine
2
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
This study is a comparative study of the characteristics of design space approximation according to various meta-models and design sensitivity evaluation using orthogonal array experiments for the Fairlead chain stopper (FCS), a detachable mooring system developed for a 10 MW-class floating offshore wind turbine. This study proposes a design methodology based on orthogonal array experiments that enables the efficient exploration of the best design case and generation of highly accurate meta-models. Thickness dimensions of the main components of the FCS are applied as design variables, while the output responses are defined as weight and strength performances. The quantitative effect of each design variable on the output responses is evaluated from the design sensitivity analysis. Various meta-models are generated, including the response surface methodology (RSM) model, Kriging, Chebyshev orthogonal polynomial (COP), and radial basis function neural network (RBFN). From the comparison of meta-model accuracy, the most appropriate meta-model is confirmed for the exploration of the FCS design space. Among the meta-models, RBFN represents accuracy for the design approximation of the FCS.
Abstract
This study is a comparative study of the characteristics of design space approximation according to various meta-models and design sensitivity evaluation using orthogonal array experiments for the Fairlead chain stopper (FCS), a detachable mooring system developed for a 10 MW-class floating offshore wind turbine. This study proposes a design methodology based on orthogonal array experiments that enables the efficient exploration of the best design case and generation of highly accurate meta-models. Thickness dimensions of the main components of the FCS are applied as design variables, while the output responses are defined as weight and strength performances. The quantitative effect of each design variable on the output responses is evaluated from the design sensitivity analysis. Various meta-models are generated, including the response surface methodology (RSM) model, Kriging, Chebyshev orthogonal polynomial (COP), and radial basis function neural network (RBFN). From the comparison of meta-model accuracy, the most appropriate meta-model is confirmed for the exploration of the FCS design space. Among the meta-models, RBFN represents accuracy for the design approximation of the FCS.