Bayes Bootstrap方法在小样本参数估计中的应用

Application of Bayes Bootstrp Method in Small Samples Parameter Estimation

针对小样本实验数据的概率分布特征较难确定,传统小样本估计方法无法提供准确的参数估计问题,工程上常用Bayes Bootstrap方法对小样本可靠性参数进行估计;鉴于该方法对具体问题有其局限性,论文对Bayes Bootstrap方法进行了改进,在不改变原样本数据的基础上将样本量进行了自助扩充,并利用Bayes Bootstrap方法对扩充后的样本进行参数估计;最后结合具体算例分析,运用蒙特卡罗仿真方法进行建模仿真,验证方法的可行性。

Because it is difficult and complex to determine the probability distribution of small samples,it is improper to use traditional probability theory to process parameter estimation for small samples.Bayes Bootstrap method is always used in the project.But,the Bayes Bootstrap method has its own limitation.In this paper,an improvement is given to the Bayes Bootstrap method.The method extends the amount of samples by numerical simulation without changing the circumstances in a small sample of the original sample.Finally,the Monte Carlo（MC）is used to model and simulate with specific small sample problems.The effectiveness and practicability of the improved-Bootstrap method is proved.