基于混淆矩阵的证据可靠性评估

在基于证据理论的融合识别系统中，对证据可靠性进行评估进而对证据进行修正是解决证据冲突的有效途径。在传感器混淆矩阵的基础上定义了后验概率向量，通过分析后验概率向量与传感器输出的证据之间的关系来对证据可靠性进行评估，充分考虑了传感器的静态可靠性与动态输出。最后将该方法运用于基于证据可靠性评估的融合识别中，算例表明该方法可以有效降低可靠性证据的影响，得到较好的融合效果。

Evaluating evidence reliability based on confusion matrix

It is of great significance to evaluate evidence reliability in data fusion based on evidence theory. Conflict in evidence can be degraded by modifying evidence. A new method for evaluating evidence reliability is proposed. The posterior probability is first obtained by confusion matrix. The reliability factor comes into being by analyzing the relation between evidence and posterior probability. Both of the static reliability and dynamic output are taken into consideration in this progress. Then this method is applied in the field of identification fusion based on evaluating evidence reliability. Numerical simulation demonstrates the good performance of the proposed method which can reduce the influence of low reliability evidence.