主观不确定性的统计分析

建立信度函数是主观不确定性分析的基础，但其统计推断方法与经典的数理统计方法存在较大差别．首先相对经典的数理统计方法，阐述了主观不确定性的母体和样本的概念，指出它们在数学上都是信度函数，调查内容则是信度函数的特征量；利用多元统计方法，给出了信度函数特征量的多元统计方法；最后，提出根据特征量拟合信度函数的方法．这些为信度函数的统计推断提供了基本的方法，也为研究工程实际中的主观不确定性奠定了一定的基础．

Statistical inference of subjective uncertainty

The establishment of belief function provides the ground work for the analysis of the subjective uncertainty, but the method of inference is by far different from that of the traditional statistics. The sample and population of subjective uncertainty are highlighted in relation with the traditional statistics in the paper, and designated that their sample and population are both belief functions in mathematics, and contents of investigation prove to be the characterized variable of belief functions. By means of multi-variables statistics, an approach to statistical inference of characterized variables of belief functions are presented. Finally, the method adopted to determine the belief functions based on characterized variables is put forwards. These provide the basic method to infer the belief functions by statistics, and establish the basis for the analysis of the subjective uncertainty in practice to a certain extent.