基于D-S理论的靶压幅度不确定性量化方法

Uncertainty quantification of target voltage amplitude using Dempster-Shafer theory

针对关键参数测试样本数有限的情况下,概率理论、区间分析等方法在对输出靶压幅度进行不确定性定量评价时存在局限性和不合理性,将D-S理论引入到靶压幅度的不确定性量化中,根据小子样测试信息得出不确定性参数的基本信任分配,以信任函数和似然函数构造靶压幅度的上下界概率分布,并以MonteCarlo方法求解。实验和仿真得出了靶压幅度的近似概率分布、置信区间及期望值分布区间等信息,并表明:与传统的概率方法相比,该方法避免了根据小样本测试信息构造概率分布的难题;与区间分析方法相比,该方法可得到更丰富的信息。

As probability theory and interval analysis have some shortages in uncertainty quantification of target voltage in the case of limited test samples of key parameters,an uncertainty quantification method based on Dempster-Shafer（D-S） theory was described.Basic belief assignments were calculated on the basis of small samples,and belief and plausibility functions were used to construct the lower and upper probability distributions of target voltage amplitude with Monte Carlo simulation.Experiments and simulations were conducted to obtain the information of target voltage amplitude,such as its approximating probabilistic distribution,confidence interval,and interval of expected values.The results show that,compared with the conventional probability method,the method avoids the problem of constructing probability distribution with small samples,and it can get more information than interval analysis.