证据理论合成规则的一点修正

Modification on Combination Rules of Evidence Theory

讨论了证据理论合成规则期望满足的基本性质．通过实例分析了证据理论合成规则中存在的问题，并提出了一种解决问题的新的融合方法．该方法能处理相互冲突的证据，并能根据证据携带的信息量决定聚焦的权重．算例表明，该方法能根据需要将证据的权重聚焦到集合中更小的子集，又能根据需要向上回到集合中更大的子集，以解决不一致证据的问题，它还能根据子集的基数大小决定聚焦的权重。

Decision level identity fusion was discussed. The Dempster Shafer's method provides for a general level of uncertainty, that is, it does not require an exhaustive set of hypotheses to be defined. However it hasn't the ability to handle inconsistent pieces of evidence and there are some problems on the way it focuses the weight of evidence. Some desirably basic features concerned with the combination rules of evidence theory were discussed. Justification of the combination rules of evidence theory was introduced in detail by examples. A new algorithm of data fusion capable of handling the existing problems was proposed. The new method is capable of handling inconsistent evidence and related evidence focusing to the amount of information resident in pieces of evidence. The examples demonstrate that basic probability mass can be focuses “downward” onto smaller subsets if so warranted, or resolving inconsistencies to larger subsets if so warranted. It focused weight downward rigorous regard to the size of subsets. The results of the new algorithm are more reasonable.