基于多超椭球模型的结构非概率不确定性分析

打开文本图片集

Structural Non-Probabilistic Uncertainty Analysis Based onMulti-super-ellipsoid Model

HUANG Ran LIU Xuekong* NI Wanjing （China Electronics Standardization Institute）

Abstract： Uncertainty problems of engineering caused byuncertain factors such as structures，materials，and loads are becoming more and more serious.How to quantify the various uncertainties and analyze the corresponding uncertainty propagation in complex computational models has atracteda lot of attention.Therefore，the paper is based on a nonprobabilisticuncertainty propagationmodel forthemulti-super-elipsoidal model.To address the problemofcognitive uncertainty caused byinsuficient structuralinformation，a multi-super-elipsoidal model isestablised toquantify the uncertainty.Furthermore，abidirectional BayesianoptimizationalgorithmbasedontheKriging modelisproposedtobalance thecomputationalefficiencyandaccuracyofuncertaintyanalysis.Thealgorithmcanactivelyselectthepointintheuncertain regionquantified bythe multi-super-ellpsoidal model toupdate the Kriging model，andsimultaneouslysolve the maximum and minimum bounds of uncertainty analysis.The analysis results of anumerical example and an enginering example validate the effectiveness of the proposed algorithm.

Keywords：uncertaintyanalysis;uncertaintyquantification;multi-super-ellipsoidalmodel;Kriging;Bayesianoptimization

0 引言

现代工程对复杂系统进行详细分析的需求日渐提升，同时装备系统的性能预测越来越依赖于数值模型，因此在系统设计时对其高性能、可靠性和功能健壮性的期望越来越高。（剩余8330字）