A New Combination Rule in Evidence Theory Based on the Strategy of Local Distribution of Conflict
Jinping Li · Journal of Jinan University · 2006
In evidence theory based on the problem of probability explanation, the combination rule is usually used in data fusion technology. But the result obtained by traditional combination rule of evidence theory is usually not in accordance with what people think when the evidences have high conflict. In order to avoid this defect, we put forward a new combination rule based the strategy of local distribution of conflict, i. e., the conflict is mainly distributed to those subsets which contribute to the conflict, and the distributed conflict value depends on the average belief value of the subset and the reliability of the evidence. Computation indicates that the improved combination rule can solve the problems caused by highly conflictive evidences as well as get more useful information from the evidences, which makes the combination result more reasonable.