克里金
数学优化
替代模型
计算机科学
模糊逻辑
约束(计算机辅助设计)
内环
最优化问题
控制理论(社会学)
数学
人工智能
控制器(灌溉)
机器学习
几何学
控制(管理)
农学
生物
出处
期刊:AIAA Journal
[American Institute of Aeronautics and Astronautics]
日期:2022-12-01
卷期号:60 (12): 6814-6824
被引量:1
摘要
Time-dependent failure possibility-based design optimization (T-PBDO) can minimize the general cost while meeting the failure possibility requirement of aircraft structure in the service life. The accuracy of the T-PBDO solution obtained by existing efficient methods may be problematic in the case of nonlinear performance functions or multiple minimum performance target points. To overcome this limitation, this paper proposes a new double-loop method based on the adaptive kriging (AK) model and fuzzy simulation (FS), referred to as DL-AK-FS, is used to efficiently solve T-PBDO. In DL-AK-FS, to replace the real constraint performance function for dealing with the time-dependent failure possibility (TDFP) constraint, the inner loop is to adaptively construct a single-loop kriging model of the constraint performance function in the FS candidate sample pool. The outer loop is to search the optimal design parameters by optimization algorithm. The kriging model is first built in an augmented space that is spanned by design parameters and fuzzy inputs, and then it is adaptively and timely updated during the optimization iteration. Moreover, the strategy of reducing the size of the FS candidate sample pool is adopted to further improve the efficiency of analyzing the inner TDFP while ensuring the accuracy of the optimization solution. The strategy of combining FS with the AK model can extend the engineering applicability of the DL-AK-FS in estimating the inner TDFP, which is not limited by the complexity of the time-dependent performance function. The optimization results show that the proposed DL-AK-FS method in this paper is efficient and accurate for solving T-PBDO.
科研通智能强力驱动
Strongly Powered by AbleSci AI