A novel dynamic weight allocation method for multisource information fusion
Yuting Li, Fuyuan Xiao · International Journal of Intelligent Systems · 2020
In multisource information fusion, the weight assignment is critical to improve the fusion performance, especially under uncertainty and conflict situations. As for uncertainty, the probability distribution is a desirable method to model uncertainty. In this context, a novel dynamic weight allocation method is first proposed, which considers three situations: nonconflict, general conflict, and high conflict. In particular, according to the degree of conflict, different weight allocations are considered by means of generalized information quality and negation operation. After that, a novel dynamic weight allocation method is proposed that contains three subalgorithms: NonC algorithm, BNC algorithm, and BIC algorithm. In addition, a novel multisource information fusion method is presented based on the newly designed dynamic weight allocation method. Besides, some numerical examples illustrate its feasibility. Finally, an application in target recognition demonstrates the practicability of the proposed fusion method.