论文部分内容阅读
鉴于传统冲突量参数无法有效地衡量证据间的相似程度,提出一种基于分类修正的多证据合成方法,以解决证据合成中的高冲突悖论和“0”悖论.首先,利用证据距离参数、冲突量参数和方向角度参数共同衡量各证据间的相似程度,将证据分为一致证据、不冲突证据、低冲突证据以及高冲突证据4类;然后,利用3个参数赋予各类证据不同的修正系数;最后,利用Dempster规则对修正后的证据进行合成.算例分析表明,所提出的方法能够较好地解决高冲突悖论和“0”悖论,而且保留了证据理论优良的数学性质.
In view of the fact that traditional conflict parameters can not effectively measure the similarity between evidences, a multi-evidence synthesis method based on classification correction is proposed to solve the high conflict paradox and “0 ” paradox in evidence synthesis.First, The parameters of distance, the parameter of conflict and the parameters of direction angle jointly measure the degree of similarity between the evidences. The evidences are divided into four categories: uniform evidence, non-conflicting evidence, low conflicting evidence and high conflicting evidence. Then, Finally, Dempster’s rule is used to synthesize the modified evidences.Examples show that the proposed method can solve the high conflict paradox and the “0” paradox well and retain the evidence theory Excellent mathematical properties.