用自助法估计外弹道测量数据随机误差分布特性

Estimation of the Random Error Distribution Characteristics of Exterior Ballistic Measurement Data Based on Bootstrap Method

针对外弹道测量数据的有限性使随机误差的分布估计不够准确的问题,提出了基于Bootstrap（自助）方法的外弹道测量数据随机误差分布特性估计。首先对测量数据采用样条分频技术得到随机误差的初步估计,然后采用三次多项式拟合修正样本经验分布函数,替换传统Bootstrap方法中经验分布函数,得到随机误差的Bootstrap样本,计算Bootstrap样本的统计量,从而得到精度较高的外弹道测量数据随机误差模型。从仿真数据验证结果来看,解决了传统Bootstrap方法仿真随机样本相对比较集中的问题。

As a result of the limitedness of ballistic measurement data,distribution estimation of the random error is not accurate enough.To solve this problem,this paper proposes random error distribution characteristics estimation of ballistic measurement data based on Bootstrap method.First,preliminary estimation of the random error is obtained with spline frequency division technology from measurement data.Then,empirical distribution function of the traditional Bootstrap method is replaced by the empirical distribution function of cubic polynomial fitting correction sample to obtain the Bootstrap sample of the random error.High accuracy random error model of the ballistic measurement data is got by computation of the statistics of the Bootstrap sample.Simulation and validation data shows that the problem of the relative concentration of the simulation random sample in the traditional Bootstrap method is solved.