基于典型样本的证据理论信度函数分配构造方法  

Approach for coustructing the confidence function distribution of evidence theory based on typical sample

作　　者：杨静[1] 田亮[1] 赵爱军[2] 刘吉臻[1] 

机构地区：[1]华北电力大学控制科学与工程学院,河北保定071003 [2]马头发电厂,河北邯郸056044

出　　处：《华北电力大学学报（自然科学版）》2008年第5期70-72,77,共4页Journal of North China Electric Power University：Natural Science Edition

基　　金：国家自然科学基金资助项目(50576022)

摘　　要：提出了一种基于典型样本的信度函数分配构造方法,先以典型样本为参照数据,计算证据在各个目标模式下的信度密度值,然后对其进行归一化处理得到信度函数分配,其中典型样本是以目标模式的特征典型值为中心的置信区间。该方法物理意义明确,具有较好的工程实用价值。This paper proposes a typical swatch-based method for constructing confidence function distribution. It takes typical sample as comparable data and computes the confidence density of each goal pattern. Then carries on normalized processing to the typical sample to obtain the confidence function distribution. The typical sample is the representative interval of goal pattern and takes the typical value as the center. This method has fine use value of project and clear significance for physics.

关 键 词：证据理论 典型样本 信度函数分配 构造 

分 类 号：TK223[动力工程及工程热物理—动力机械及工程]