Abstract: Conflict of evidence is one of the most important factor which leads to the fusion result of evidence theory unsatisfactory. Thus the evidence conflict has been the key issue to be solved in evidence theory. By using a linear combination of ambiguity measure, discord measure and nonspecificity measure, a new uncertainty measurement method of evidence is presented. The new method covers uncertainty information more fully of discord and nonspecificity which are included in the body of evidence. This method makes the information more complete in the result of uncertainty measure. Based on the new method, an exponential function is used to construct weights, and the body of evidence is preprocessed. Finally, Dempster-shafer theory, proportional conflict redistribution 2 and proportional conflict redistribution 5 are used to fuse the preprocessing evidences. Numerical examples show that the fusion results are reasonable, and then the new uncertainty measurement method is effective.

Key words: uncertainty measure, evidence theory, ambiguity measure, discord, nonspecificity, information fusion

摘要： 证据的冲突是导致证据理论融合结果不理想的重要因素，对证据冲突的处理一直是证据理论中要解决的关键问题。通过对多义度、不一致度和非特异度进行线性组合，提出一种新的证据不确定性度量方法。新方法能够更全面地涵盖证据体中所包含的不一致和非特异性两类不确定信息，使得证据不确定性度量结果所包含的信息更为完整。在此基础上，采用指数函数构造权重，对证据体进行预处理，然后采用DST、PCR2和PCR5进行融合。算例结果表明融合结果合理，新的证据不确定性度量方法有效。

关键词: 不确定度, 证据理论, 多义度, 冲突, 非特异性, 信息融合