基于概率分布距离的多响应模型确认度量

Multiple response model validation metric based on distance of probability distribution

针对模型确认中的确认度量问题,构造实验观测数据经验概率分布的置信包络.通过计算其与模型响应概率分布之间距离的上/下确界,给出基于概率分布距离确认度量的置信区间.通过构造与实验观测数据有关的协方差矩阵,给出基于概率分布距离的多响应模型确认度量及其置信区间的求解方式.该度量利用了模型输出与实验观测的完整概率分布信息,并且考虑了各模型响应间的相关性.算例仿真结果表明其确认错误率低于现有的其他两种确认度量.

For the metric for model validation, the confidence interval of the validation metric is presented by calculating the infimum and supremum of the distance between the empirical probability distribution of model response and the confidence interval of the experimental distribution. Besides, a multiple response validation metric based on the distance of probability distribution is proposed with its confidence interval by constructing experimental covariance matrix. The metric makes use of the whole data distribution and considers the correlation among multiple model responses. Simulation results show that the validation error rate of the proposed metric is lower than other two metrics.