一种改进的有效冲突证据融合方法

One Advanced Efficient Fusion Approachof Conflict Evidence

针对经典证据理论Dempster规则无法有效合成冲突证据的问题,结合现有改进规则方法和改进模型两类方法的优势,提出一种改进的冲突证据融合处理方法。引入证据平均值的概念,通过计算各证据与证据平均值之间的距离来确定可信度作为权重,以区分各证据间的相关性和差异;再加权平均进行合成获得融合结果;并通过比较验证了该方法不仅有效,且在收敛速度和可靠性上更加优越。

The dempster rule cannot effectively combine the conflict evidence. The existing advanced rules and advanced models were analyzed and combined, and one advanced efficient fusion approach of conflict evidence was proposed. In order to distinguish the correlations and differences of the evidences, the average evidence distance was introduced, and every evidence support degree was determined through calculating the distance between every evidence and the average evidence value as the weight, then weighted mean combined to get the fusion results. The approach was compared to verify it was not only valid, but also superior than the others in the convergence speed and reliability.