Detection of Influential Uncertain Parameters in Monte Carlo Evaluation
Detection of Influential Uncertain Parameters in Monte Carlo Evaluation
复制标题
蒙特卡罗评估中影响不确定参数的检测
DOI：
10.2322/jjsass.55.117
复制
发表时间：
2007
期刊：
影响因子：
--
通讯作者：
Toshikazu Motoda
中科院分区：
文献类型：
--
作者：
Toshikazu Motoda
在航空航天项目中，飞行前评估对任务的成功至关重要，因为在真实环境中进行飞行测试通常是困难的或不可能的。蒙特卡罗仿真是飞行前评估的有力工具，因为包含各种不确定参数的非线性系统可以直接进行评估。随着计算机能力的提高，蒙特卡罗模拟已在世界各地的各种航空航天项目中得到应用。在系统评估后，重要的是要检测出影响系统性能下降的不确定参数，以便研究改进系统的措施。然而，在蒙特卡罗模拟中，由于各种不确定参数同时包含，且其大小是随机产生的，因此检测这些参数往往很困难。两个以上的不确定参数的相互作用可能会受到影响。本文提出了一种利用蒙特卡罗结果的统计检验来检测有影响的不确定参数的简单方法。
In aerospace projects preflight evaluation is crucial for mission success, because flight testing in a real environment is often difficult or impossible. Monte Carlo simulation is a powerful tool for the preflight evaluation because a nonlinear system incorporating various uncertain parameters can be evaluated directly. Monte Carlo simulation has been used in various aerospace projects throughout the world as the computer power increases. After the system evaluation, it is important to detect influential uncertain parameters which cause significant performance degradation so that measures for the system improvement can be studied. However, detecting those parameters is often uneasy because various uncertain parameters are incorporated simultaneously and their magnitudes are randomly generated in Monte Carlo simulation. An interaction of more than two uncertain parameters might be affected. This paper presents a simple approach to detect influential uncertain parameters applying a statistical test to the Monte Carlo results.