一种基于概率语言偏好关系的DS证据理论改进算法

An Improved DS Evidence Theory Based on Probabilistic Language Preference Relation

DS(Dempster-Shafer)证据理论在不确定信息融合等领域有着十分广泛的应用,但当出现高冲突证据时,融合结果往往有悖常理.为了解决高冲突证据合成悖论的问题,本文提出了一种基于概率语言偏好关系的DS证据理论改进算法.该方法用概率语言偏好关系既解决了专家偏好产生的困难,又准确识别了冲突专家,而且提出了冲突证据的调整方法,在保持Dempster合成规则不变的情况下,改进了融合结果.最后分别应用数值分析和案例模拟分析检验了提出方法的科学有效性和应用可行性.

DS(Dempster-Shafer) evidence theory has a wide range of applications in the field of uncertain information fusion, but when there is evidence of serious conflict, the fusion results are often counterintuitive. In order to solve the problem of high conflict evidence synthesis paradox, this paper proposes an improved DS evidence theory algorithm based on probabilistic language preference relationship. The method uses the probabilistic language preference relationship to solve the difficulties caused by the expert preference, accurately identifies the conflict experts, and proposes the adjustment method of the conflict evidence, and improves the fusion result while keeping the Dempster synthesis rules unchanged. Finally, numerical analysis and case simulation analysis are applied to verify the scientific validity and application feasibility of the proposed method.