论文部分内容阅读
针对Borgonovo基本输入变量重要度中概率密度函数积分不易求解的缺陷,基于重要度定义将范数的概念引入灵敏度分析中的重要度排序分析领域,提出一种新的基本输入变量重要度排序的计算策略,并为该计算策略选择了若干等价范数,从而避免求解积分的困难.该方法将所求积分用其等价范数进行替代,同时采用适用性更加广泛的一种正则化基本输入变量重要度的计算方法.从理论上讲,该计算策略能够为重要度排序计算提供多种等价范数形式.从所提计算策略与Borgonovo方法、Liu方法的比较可以看出:所提方法是三种方法中最简单有效的.最后,给出两个算例来说明所提方法的可行性.
Aiming at the defect that the integral of probability density function in Borgonovo’s basic input variables is not easy to solve, the concept of norm is introduced into the field of importance ranking analysis in sensitivity analysis based on the definition of importance, and a new algorithm for calculating the importance of basic input variables Strategy and chooses some equivalent norm for the calculation strategy so as to avoid the difficulty of solving the integral.The method replaces the integral with its equivalent norm and adopts a more generalized regularization basic input In theory, this calculation method can provide multiple equivalent norm forms for the calculation of importance rankings.Comparing with the Borgonovo method and Liu method, we can see that the proposed method Is the simplest and effective of the three methods.Finally, two examples are given to illustrate the feasibility of the proposed method.