基于自助法仿真的不确定度抽样反分析方法研究  

RESEARCH ON UNCERTAINTY RE-SAMPLING REVERSE ANALYSIS METHOD BASED ON BOOTSTRAP SIMULATION

作　　者：祖先锋[1] 毛健人 蒋志文[3] 

机构地区：[1]空军第一航空学院,河南信阳464000 [2]北京临近空间飞行器系统工程研究所,北京100076 [3]国防科学技术大学计算机学院,湖南长沙410073

出　　处：《计算机应用与软件》2011年第6期215-217,256,共4页Computer Applications and Software

摘　　要：针对测量链中激励源信号不确定度评定难的问题,借助于反分析思想和蒙特卡洛仿真分布传播原理,提出了不确定度分布抽样反分析法。通过Bootstrap自助法再抽样扩展样本容量,结合小样本容量的概率密度函数估计及其不确定度计算,由输出量仿真出输入量概率分布,实现了蒙特卡洛仿真分布传播的逆向过程,间接完成由测试信号到源信号的不确定度评定。实验与仿真结果表明,由反分析法所得不确定度参数与实际检定结果接近,从而验证了不确定度分布抽样反分析法的有效性和可行性。To tackle with the difficulty in uncertainty evaluation of source signals within a measurement chain,depending on the reverse analysis idea and Monte Carlo simulation distribution and propagation principle,an uncertainty distribution re-sampling reverse analysis method is proposed.The method expands the sample capacity by Bootstrap re-sampling,integrates small sample capacity probability density function estimation along with its uncertainty computation,simulate input value probability distribution with output values,realizes Monte Carlo simulation distribution reverse process,and indirectly accomplishes the uncertainty evaluation from testing signals to source signals.Experiment and simulation results show that the uncertainty parameters calculated by reverse analysis method are close to actual detection results;therefore the effectiveness and feasibility of uncertainty distribution re-sampling reverse analysis method is validated.

关 键 词：测量不确定度 反分析法 自助法再抽样 蒙特卡洛仿真 

分 类 号：O211.67[理学—概率论与数理统计]