中文摘要：

针对证据理论Dempster Sharer（D—S）组合规则存在的不足，考虑辨识框架的非完备性，提出一种基于模糊聚类分析的混合证据组合方法。该方法利用Pignistic概率距离构建证据间的模糊相似矩阵及聚类有效性指标，通过传递闭包法对证据进行最优聚类及可信度求取，对证据进行修正后，采用D—S规则对同一类中的证据组合，利用统一信度函数组合方法对不同类中的证据进行最终合成。仿真实验表明了所提方法的合理性和有效性。

英文摘要：

To deal with the shortcomings of the Dempster Shafer （D-S） combination rule, a hybrid evidence combination method is proposed based on fuzzy clustering analysis and the incompleteness of a discernment frame. The method formulates the fuzzy similar matrix and designs the clustering validity index by using the Pignistic probability distance. The evidence is clustered optimally and its reliability is calculated by the fuzzy transitive closure method, the original evidence is modified by its reliability, then the evidence in the same set is combined with D-S rule of combination and the evidence in different sets is combined with the combination method of the generic belief function. Some computational simulations and comparisons show the efficiency and rationa lity of the proposed approach.