网络多信道多源信息融合技术的改进研究

由于D-S融合算法采用的证据合成规则无法有效处理证据冲突问题,并且BP神经网络融合算法在样本波动情况下需要再次训练完成融合,容易出现局部最小值问题。因此,融合D-S融合算法和BP神经网络融合算法,提出基于上下文权值的多信道多源信息复合融合算法,其先基于检测数据的上下文,采用D-S融合算法和BP神经网络融合算法,对对应上下文内的数据进行融合处理,获取的融合结果被设置相应的权值,再将多个并行融合结果进行加权汇总,得到最终的融合结果,并同设置的阀值实施对比,获取最终的判决结果。实验结果说明,所提算法可有效处理证据冲突的融合问题,具有较高的准确性,融合效果佳。

Research on improvement of multichannel multisource information fusion technology in network

Because the rule of evidence combination used by D-S fusion algorithm can not effectively deal with evidence conflict, and BP neural network fusion algorithm in the sample fluctuations needs to train again for completing the fusion, it is easy to cause local minimum problem. Therefore, we integrate D-S fusion algorithm and BP neural network fusion algorithm to propose multichannel multi source data fusion algorithm based on context weight. Firstly, this paper used D-S fusion algorithm and BP neural network fusion algorithm based on the context of detected data for the fusion processing of corresponding data in context, then the obtained fusion result was set for the corresponding weights, multiple parallel fusion results was weighted to get final fusion results. We compared with the setting threshold to obtain the final verdict. Experimental results show that the proposed algorithm can effectively deal with the fusion problem of evidence conflict, which has high accuracy and good fusion effect.