摘要
现代仿真试验设计面临着试验影响因素数目众多以及因素之间关系复杂的问题,为了简化最终的试验设计,需要对众多试验涉及的因素进行辨识。全局分析比局部分析更能反映参数取值以及参数之间的关系对结果的影响,通过比较目前最常用的全局灵敏度分析方法的优劣,最终选择Morris方法来进行参数辨识的试验设计。由于Morris方法针对非线性问题有某些局限性,根据需求创造性地对Morris方法进行了改进。为了保证试验能够跨平台进行,利用MFC与可扩展标记语言(XML)实现Morris方法,完成试验设计工具,为仿真试验设计提供一种适用性较强的工具。
The modern design of simulation experiment needs to identify so many factors. For easy design of the final experiment, the important factors should be selected. The one-factor simulation experiment is less useful for analysis of validating the importance of model parameters and the relationship between parameters. This paper analyzes several overall sensitivity analysis methods, and the Morris Method is selected. This paper uses Morris Method to make the scenarios, and according to the real condition this method is developed. For cross-platform experiment, the tool using the Morris Method is developed by MFC and Extensible Markup Language （XML）, and it provides an applicable tool for design of simulation experiment.
出处
《系统仿真技术》
2014年第2期171-176,共6页
System Simulation Technology
关键词
Morris方法
灵敏度分析
仿真试验
试验设计
可扩展标记语言
Morris method
sensitivity analysis
simulation experiment
design of experiment
Extensible Markup Language （XML）
