一种DSm理论相关证据模型  

Model for Dependent Evidences in DSmT Framework

作　　者：王进[1] 孙怀江[1] 

机构地区：[1]南京理工大学计算机系,南京210094

出　　处：《计算机科学》2009年第8期260-263,295,共5页Computer Science

摘　　要：DSm理论(Dezert-Smarandache Theory,DSmT)作为D-S(Dempster-Shafer)证据理论的推广,在进行多个证据合成时同样要求证据相互独立,这是一个非常苛刻的条件,在实际中很难满足。提出了在DSm理论框架中表示相关证据的一种模型,其中两个相关证据各由一个独立源证据和一个相关源证据正交和合成,相关证据的合成就归结为这两个独立源证据和这个相关源证据的正交和合成。为此就要先辨识出独立源证据。证明了在自由DSm模型中辨识结果的唯一条件,并针对条件不满足的情况提出了一种近似的解决方法。DSmT（I）ezert-Smarandache Theory） which is an extension of Dempster-Shafer theory assumes that sources of evidences are independent as the same as l）empster-Shafer Evidence Theory. This is a very strict condition that is difficult to satisfy in practice. A new model used to represent interrelated evidences in DSmT framework was proposed. In this model,two dependent evidences were considered resulted from orthogonal sum of one dependent original evidence and two independent original evidences, respectively. Combining the two dependent evidences can be reduced to an orthogonal sum of the two independent original evidences and the dependent original evidence. For this, independent original evidences must be identified firstly. A condition was proven to guarantee that the identified dependent original evidence is unique in free DSmT model, and if this condition is not satisfied, an approximate solution is proposed.

关 键 词：DEZERT-SMARANDACHE理论 不确定性 相关性 

分 类 号：TP301[自动化与计算机技术—计算机系统结构] TP393[自动化与计算机技术—计算机科学与技术]