一种基于马尔科夫链的冲突证据组合方法  

A Method of Conflictive Evidence Combination Based on the Markov Chain

作　　者：李新德[1] 董清泉 王丰羽 雒超民 

机构地区：[1]东南大学自动化学院复杂工程测量与控制教育部重点实验室,南京210096 [2]底特律大学电子与计算机工程系密歇根美国

出　　处：《自动化学报》2015年第5期914-927,共14页Acta Automatica Sinica

基　　金：国家自然科学基金(60804063,61175091);航空基金(20140169002);江苏省“青蓝工程”资助计划;江苏省“六大高峰人才”资助计划资助~~

摘　　要：针对智能信息处理中Dempster组合规则不能处理高度冲突的问题,考虑到序贯证据的序列性具有高效的抗干扰性能,因此本文提出了一种基于马尔科夫链的冲突证据组合方法.首先,从经典马尔科夫链中的确定性状态描述扩展到不确定性状态描述;然后,以滑动窗口宽度l对序贯历史证据进行采样,并利用相似性测度计算的权重来修正它们,从而对修正后的历史证据进行马尔科夫建模,并根据转移概率矩阵,计算证据代表;最后,利用Murphy组合规则对该证据代表组合l-1次.当然,本文方法也比较适合批量同步融合.大量的仿真实验对比分析表明,该方法优势比较明显,有效地解决了冲突证据合成出现的问题,并能有效兼顾合成结果的鲁棒性和灵敏性.Aiming at the problem that highly conflictive evidence can not be processed by Dempster rule in intelligent information processing, a method of conflictive evidence combination based on Markov chain is proposed by considering the high-e？ciency anti-interference performance for the sequentiality of sequential evidences. At first, the deterministic state description in the classic Markov chain is extended to nondeterministic state description. And then, the past evidences are sampled sequentially according to the sliding window width l, which could be amended according to the weight computed by utilizing the similarity measure. A Markov model is established on these past evidences amended so that a transition probability matrix could be obtained, which is used to compute the evidential representative. Finally, this representative is combined with itself for l？1 times according to the Murphy0s combination method. Of course, this method also fits parallel fuse in a step. Through simulation experiments, the comparisive analysis show that the new method0s advantage is obvious. That is to say, it e？ciently solves the problem of the combination of conflictive evidences; moreover, it keeps robustness and sensibility of combinational result.

关 键 词：证据推理 冲突 马尔科夫链 状态不确定性 组合规则 

分 类 号：TP202[自动化与计算机技术—检测技术与自动化装置]