证据理论的并行计算方法

A PARALLEL COMPUTING METHOD OF EVIDENCE THEORY

针对证据理论在应用中计算量大的问题,给出了证据理论的并行计算方法模型。该模型分为3层:输入层、融合层、输出层,输入层接受证据的输入,融合层完成两个证据的融合,输出层完成证据融合之后的归一化输出。该模型的融合结果与Dempster方法的结果相一致,并保持了Dempster合成规则中的"交换律"。实验表明本文所给出的模型能够提高证据理论的融合速度。

A parallel computing method model of evidence theory is proposed to deal with the problem of large amount of calculation of evidence theory in application. The model is constructed of three layers:the input layer, fusion layer and output layer. The evidences are inputted to the input layer, fused in the fusion layer, and the fusion results are normalized and outputted by means of the output layer. The fusion results of the model are consistent with the results of Dempster synthesis rule, and the"commutative law" of Demp-ster synthesis rule is maintained in this model. Experiments show that the model presented in this paper can improve the fusion speed of Evidence theory.