考虑多场耦合的大型海上风机支撑结构可靠性优化设计方法

Reliability-based design optimization (RBDO) of offshore wind turbine support structures (OWTSS) needs to consider the air-structure-water -servo-soil fully coupling effects subject to wind, waves and currents, as well as the high-dimension and multisource stochastic uncertainties and epistemic uncertainties involved in external environment, material behavior, computational modeling and other aspects. So, the traditional probability theory is found inapplicable to the case. In this project, hybrid model of probability and probability-box (p-box) will be used to RBDO of OWTSS，and two key problems of the reliability modeling and the huge computational cost originated from multi-physical field coupling will be investigated. The first step is to establish the stochastic airfoil model and joint probability distribution function of wind-wave based on Copula function，to compute extreme load distribution of support structure based on the air-structure-servo-water fully coupled dynamics model，and to acquire the envelop boundaries of extreme load distribution by probability boundary analysis. The second step is to present the method converting the envelop boundaries to p-box, and the reliability computational approach based on active learning surrogate model. Lastly, the practical and efficient RBDO method will be researched.This project will provide the theoretical support for structural design of offshore wind turbine and offer a new calculation method for RBDO problem considering multi-physical field coupling.

海上风机支撑结构可靠性优化设计不仅需要考虑风、浪、流联合作用下的结构与气动、水动、岩土和控制的耦合作用，还需要考虑环境、材料和计算模型等方面存在的高维、多源随机不确定性和认知不确定性，因此传统概率方法难以适用。本项目拟采用概率-概率盒混合模型对海上风机支撑结构进行可靠性优化设计，围绕多场耦合下不确定建模困难和可靠性优化设计效率低两个关键问题开展研究。首先拟建立叶片翼型随机模型和基于Copula函数的风浪联合模型，利用气-结构-伺服-水全耦合动力学仿真模型计算支撑结构极值载荷分布，并运用概率边界法获取极值载荷分布的包络边界；然后提出包络边界转化为概率盒的方法和基于主动学习代理模型的可靠度计算方法；最后研究高效的可靠性优化设计方法。项目的完成将为海上风机结构设计提供理论支撑，也为复杂物理场的结构可靠性优化提供一种新的研究思路。