参数估计的数值解析法  

Numerical Analytical Method for Parameters Estimate

作　　者：宋占勋[1] 谢基龙[1] 周素霞[1] 杨月[1] 

机构地区：[1]北京交通大学机械与电子控制工程学院,北京100044

出　　处：《北京交通大学学报》2010年第1期137-139,144,共4页JOURNAL OF BEIJING JIAOTONG UNIVERSITY

基　　金：国家"973"计划项目资助(2007CB714705);铁道部科技研究开发计划重点项目资助(2006J017)

摘　　要：对目前参数估计的方法进行了总结与对比.针对不完整数据参数估计时的情况,分析了目前关于参数估计方法存在的问题.利用Matlab求解非线性方程,提出了一种高精度数值解析法.利用少量的实验数据便可求出一组参数的解,然后排列组合试验所得到的数据,可以求出多组参数的解,求解多组参数的平均值,从而得到参数的精确解.并以两种工程中常用的拟合函数正态分布和威布尔分布为例,列出了数值解析法的具体求解步骤,通过与极大似然法的比较,验证了本文方法的可行性.The current parameter estimate methods are summarized and compared. Considering data in- completion of parameter estimate, the problems existing in current parameter estimate methods are an- alyzed. One high accuracy method on parameter solving is put forward by using Matlab to solve nonlinear equation. A small quantity of experiment data can get a set of parameter solutions. The experiment data is organized and combined to get several parameter solutions, and their average value of parameter, then, the precise parameter solutions are deduced. Two kinds of function fitting that are popular for the engineering weibull distribution and normal distribution are taken as examples, and the parameter solution methods are listed. Compared with maximum likelihood, the feasibility of these methods are confirmed.

关 键 词：参数估计 可靠性 不完整数据 

分 类 号：TB11[理学—数学]