计算机工程与应用 ›› 2014, Vol. 50 ›› Issue (3): 103-107.

• 数据库、数据挖掘、机器学习 • 上一篇    下一篇

基于加权证据距离的高度冲突证据组合方法

刘志成,乔  慧,何佳洲   

  1. 江苏自动化研究所,江苏 连云港 222006
  • 出版日期:2014-02-01 发布日期:2014-01-26

Combination approach of highly conflict evidence based on weighted distance of evidence

LIU Zhicheng, QIAO Hui, HE Jiazhou   

  1. Jiangsu Automation Research Institute, Lianyungang, Jiangsu 222006, China
  • Online:2014-02-01 Published:2014-01-26

摘要: 为了有效融合高度冲突的证据,以Murphy方法和邓勇加权平均法为基础,提出了一种新的基于加权证据距离的证据组合方法。用Murphy方法确定各证据体的权重,采用修正的City Block距离加权平均求证据间的两两距离,进而获取各证据被其他所有证据支持的程度,归一化各证据的总支持度作为该证据的权重,对多源证据加权平均后再利用Dempster组合规则实现信息融合。实验结果表明,该方法能够更加有效快速地识别出目标,拥有更快的收敛速度。

关键词: 高度冲突证据, 加权证据距离, Dempster组合规则, 信息融合

Abstract: In order to efficiently combine highly conflict evidence, this paper proposes a new combination method based on the weighted distance of evidence by using Murphy’s method and Deng’s weighted averaging method. The weight of each body of evidence is determined by using Murphy’s method. The weighted distance between every two evidences is calculated under the modified City Block distance norm, so the support degree of each evidence supported by other evidences can be obtained. After normalizing the general support degree of each evidence as the weight of each evidence, the information fusion process under the Dempster combination rule can be realized. Simulation results show that the new method converges at a faster rate and can recognise the target more effectively and fleetly.

Key words: highly conflict evidence, weighted distance of evidence, Dempster combination rule, information fusion