多项式输出中相关变量的结构贡献和相关贡献分析  

Analysis of structural and correlative contributions by correlated variables in polynomial output

作　　者：吴丹清 吕震宙[1] 胡吉祥[1] 

机构地区：[1]西北工业大学航空学院,陕西西安710072

出　　处：《国防科技大学学报》2014年第5期26-32,共7页Journal of National University of Defense Technology

基　　金：国家自然科学基金资助项目(51175425);航空基金资助项目(2011ZA53015);博士学科点专项科研基金资助项目(20116102110003)

摘　　要：以二次不含交叉项的多项式为例,解析推导了正态相关输入变量对输出响应量方差贡献的结构贡献部分和相关贡献部分。通过算例验证了解析结果的正确性。将所研究的指标和已有的进行对比,归纳出结构贡献部分和相关贡献部分的侧重与统一。解析解可直接用于极限状态函数不超过两次且不含交叉项的结构和相关贡献的识别,为其他新的数值算法提供了参考对照。Based on the example of the quadratic polynomial without cross terms outputs,the structural and correlative contributions of the correlated normal input variables were derived out,which had impact on the variance contributions of output response variables.Numerical results of several examples validate the correctness of analytical solutions.Comparing the presented indices with the existent indices,some general conclusions were drawn from the structural contributions and the correlative contributions,and the differences between them were identified.The proposed analytical solutions can be directly used to identify the structural contributions and the correlative contributions of the quadratic polynomial without cross terms,which offer a reference to other numerical algorithms.

关 键 词：结构贡献 相关贡献 二次不含交叉项 解析解 

分 类 号：TB114.3[理学—概率论与数理统计]