基于修正权值混合验后的导弹精度融合评估  

Synthetic evaluation of missile precision based on mixed posteriors with modified weights

作　　者：闫志强[1,2] 蒋英杰[1] 宫二玲[1] 谢红卫[1] 

机构地区：[1]国防科学技术大学机电工程与自动化学院,湖南长沙410073 [2]二炮装备研究院,北京100085

出　　处：《系统工程与电子技术》2011年第3期712-716,共5页Systems Engineering and Electronics

摘　　要：当利用仿真实验和飞行试验两类信息进行导弹精度融合评估时,对于正态分布参数,如果将有信息验前分布与无信息验前分布进行混合,将导致混合验后分布权重无法真实反映两类数据的相容性水平,从而导致估计结果产生偏差。针对这一问题,提出了飞行试验样本所对应的有验前样本容量约束的第二类极大似然(prior samplesize constrained maximum likelihood-II,SCML-II)估计,定义了有验前样本容量约束的边际密度(prior sample sizeconstrained marginal density,SCMD)函数值。采用现场样本的SCMD替换无信息验前分布条件下的现场样本边际分布,从而修改了加权混合验后分布的融合权重。最后,通过实例分析了改进方法的估计性能。In the case of missile precision synthetic evaluation by the information of simulation test data and fly test data,for the normal distribution parameters,the informative prior is usually mixed with non-informative prior,and the mixed posteriors weights can not reflect the consistency of the two sorts of data,which brings inaccurate results.In view of this,a prior sample size constrained maximum likelihood-Ⅱ（SCML-Ⅱ） estimation is presented,and a prior sample size constrained marginal density（SCMD） is defined.The marginal of the field sample upon non-informative prior is replaced by the SCMD of the field sample,which brings the modification of weights of mixed posteriors.Finally,the estimation performance of the modified method is analyzed by an example.

关 键 词：精度分析 混合验后分布 仿真可信度 融合评估 第二类极大似然 

分 类 号：O212.8[理学—概率论与数理统计] TJ760.6[理学—数学]