仿真输出分析中置信区间的Bootstrap估计方法

Bootstrap Estimators of Confidence Intervals in Simulation Output Analysis

介绍了一种非参数统计分析方法，并将此方法应用于仿真输出分析，从而解决了传统方法的局限性。建立了感兴趣的性能测度的Ｂｏｏｔｓｔｒａｐ置信区间估计的步骤，并通过仿真实验进行了验证。Ｍ／Ｍ／１排队系统的仿真实验结果表明，这种方法能给出合理的点估计和有效的置信区间，而且仿真精度较传统方法有明显提高。

Discusses a new nonparametric statistical method and applies it to simulation output analysis.This can be used to solve the limitations of the traditional methods.The step of making Bootstrap interval estimation is given.The simulation results for M/M/1 queue show that the Bootstrap method can produce the valid confidence intervals and improve the simulation precision obviously.