Solution of Evidence Distance in Multi-Source Information Fusion

By relative study on distance measure and conflict evidence synthesis of evidence theory, we find it is very difficult to define strict evidence distance measurement directly. While indirect evidence distance measuring which is based on BPA probability transformation is simple and effective. Current BPA probability transformation methods lack reasonable transformation basis and strong subjective reliability. In order to keep transformation probability of original BPA, we choose non-specificity as uncertain weight to establish non-specificity-based probability transformation method for BPA. Besides, for the contradiction of calculation between the most specific BPA and probability transformation, a combined solution method of “hypothesis-verification” between these two parts is provided. Experimental results show that the conservative degree of proposed probability transformation process is proportional to uncertainty degree of BPA, which proves the improved rationality of project. DS combination method is also used to weighted the synthesis of obtaining fusion results and its effectiveness is proved by simulation cases.