摘要
两阶段序贯分析方法是仿真(模拟)实验研究中的一种常用方法,它虽然可以通过控制实验给出符合精度要求的结果数据,但以其现有形式有时却无法避免冗余计算的存在和实验过早终止的可能。在回顾了现有终态仿真解算终止策略特点的基础上,分析了所需样本总数的估计值与当前可用样本间的变化关系,提出了一种基于参数稳态判定的改进两阶段序贯方法。该方法增加了策略实施的主动性,可以在所需采样样本总数量n值较大的情况下,有效改善两阶段方法中对n值的初始估计,从而改进计算效率并能保证解算结果的准确性。
Two-stage sequential stopping procedure is used for terminating simulation analysis.It can control precision of result data that is required,but cannot avoid redundancy computation and early stopping.Firstly,the existing stopping procedures of terminating simulation were reviewed;then,the relationship between the quantity estimating of the samples total needed and the current amount of the samples usable was analyzed;finally,a modified method based on parameter steady-state estimate was proposed.This method can improve estimating of n(samples total needed) if the n is biggish,and ensure the validity of the simulation result.It can increase go-aheadism for the procedure actualizing.
出处
《国防科技大学学报》
EI
CAS
CSCD
北大核心
2011年第1期120-124,共5页
Journal of National University of Defense Technology
基金
国家自然科学基金资助项目(60974073
60974074)
国家部委基金资助项目(9140C640505)
关键词
决策分析
建模仿真
序贯分析
参数稳态判定
敏捷性
有效性
decision-making
modeling and simulation
sequential analysis
parameter steady-state estimate
agility
validity