区间信度结构下的证据合成方法研究

摘要

In the process of data fusion with evidence theory,precise belief structures required in original evidence theory are not always available.At this point,approaches for combination of interval-valued belief structures is needed.Exist-ing approaches in DST and DSmT framework are reviewed,examined and analyzed.The paper proposed four version of CDI (Distribution Conflict of Interval-valued belief structures)for combination of evidence.CDI1 ~CDI4 are all fit for DST and DSmT framework.The approaches can deal with imprecise and uncertain information.The accuracy of combination result from CDI1 ~4 improves gradually.Numerical examples and comparison of results of existing approaches for interval-valued belief structure are provided throughout the paper.%在使用证据理论进行数据融合的过程中，有时精确的信度结构很难获得，此时需要对区间信度进行合成．本文分析了在DST和DSmT框架下的区间证据合成问题，对目前使用的方法进行了简要的回顾．通过对优化方法进行研究，提出了4种应用于区间信度组合的优化方法．CDI1～CDI4方法都可应用于DST和DSmT框架，对不精确，不确定以及冲突的信息进行合成，合成结果准确度逐步提高．文章最后给出了算例验证，并与其他区间信度合成方法的进行对比．