改进的D-S证据理论算法

An improved D-S evidence theory algorithm

针对DS证据理论存在的诸如一票否决、证据冲突过大的问题,提出了一种改进方法。该方法利用指数函数对证据进行重新定义,有效避免了零焦元元素对融合结果的影响,引入关联系数的概念解决冲突系数不能够有效反映证据之间冲突关系的情况。利用关联系数矩阵对焦元元素的关联程度作为权重对原证据进行加权,并利用DS证据理论合成公式对加权后的数据进行融合。通过仿真分析了调节因子的取值范围,并通过与其他算法进行比较,得出新算法在解决一票否决问题方面的优越性。

Aiming at one ticket veto and too much conflict evidence existing in DS evidence theory, an improved method is proposed. The method redefines the evidence though exponential function and it avoids the influence caused by focal element on fusion result. The situation that conflict coefficient can not effectively reflect the conflict relationship between evidences is solved by introducing the definition of correlation coefficient. The original evidences are weighed by the connection degree of focal elements using correlation coefficient matrix and fuse the weighed data by DS evidence theory. The range of regulato and the ry factor is analyzed though the simulation. The proposed method is compared to other methods experimental result shows that the proposed method has advantage in one ticket veto.