Dempster's rule combination is invalid when the results of recognition are highly conflicting ,the shortcom‐ings of some solutions are analyzed in this paper .A new method is proposed to handle this problem .The new method firstly determine the weighting coefficient by the combination of pignistic probability distance of evidence and clarity measure ,sec‐ondly amends the evidence with the weighting coefficient ,lastly combines the average evidence by Dempster rule .The simu‐lated experiment indicates that the new method can not only obtain a better result ,but also accelerate convergence .%针对高度冲突识别信息 Dempster 规则融合失效问题,分析了有关改进方法及其存在的不足,提出了一种新的证据组合方法。新方法首先根据证据之间 pignistic 概率距离和基于信息熵的证据清晰度综合确定各证据权重系数,然后使用该权重系数对证据源进行修正,最后将修正后的加权平均证据进行 Dempster 规则组合。仿真实验结果表明新方法在目标综合识别中合成结果更好,收敛速度更快。
展开▼