本文選題：多輸出 + 重要性測度�。� 參考：《南京航空航天大學(xué)學(xué)報》2017年03期

【摘要】：基本變量重要性測度分析是結(jié)構(gòu)安全評估以及工程優(yōu)化設(shè)計的一項必要工作。本文結(jié)合矩獨立重要性測度思想和基于方差的重要性測度思想,提出了一種新的適用于多個輸出的重要性測度。所提測度利用概率積分轉(zhuǎn)化(Probability integration transformation,PIT),將輸出的不確定性由輸出的聯(lián)合分布函數(shù)來表征。在多輸出情況下,相比單純基于方差的重要性測度,所提測度能夠同時包含各個輸出的不確定性及其相關(guān)性。而相比單純的矩獨立重要性測度,由于它使用方差來衡量變異性,因而求解過程更為簡便。在計算多輸出的分布函數(shù)時,本文采用基于分數(shù)矩的極大熵方法結(jié)合Nataf變換法,在保證求解精度的同時大大降低了模型調(diào)用次數(shù)。數(shù)值和工程算例說明了該測度的合理性以及計算方法的高效性。
[Abstract]:The importance analysis of basic variables is a necessary work in structural safety assessment and engineering optimization design. Combining the idea of moment independent importance measure and variance based importance measure, a new importance measure suitable for multiple outputs is proposed in this paper. The proposed measure uses probabilistic integration transformation to characterize the uncertainty of the output by the joint distribution function of the output. In the case of multi-output, the proposed measure can include the uncertainty and correlation of each output at the same time, compared with the simple importance measure based on variance. Compared with the simple moment independent importance measure, it is easier to solve because it uses variance to measure variability. When calculating the distribution function of multiple outputs, the maximum entropy method based on fractional moments combined with Nataf transform method is used in this paper, which ensures the accuracy of the solution and greatly reduces the number of model calls. Numerical and engineering examples show the rationality of the measure and the efficiency of the calculation method.
【作者單位】： 西北工業(yè)大學(xué)航空學(xué)院;
【基金】：國家自然科學(xué)基金(51475370)資助項目 中央高校基本科研業(yè)務(wù)費(3102015BJ(Ⅱ)CG009)資助項目
【分類號】：TB114
，

本文編號：1951038