汪晓钢, 钱文瀚. 分布式多传感器决策融合的新方法──双端检验法[J]. 机器人, 1995, 17(3): 129-139.
引用本文: 汪晓钢, 钱文瀚. 分布式多传感器决策融合的新方法──双端检验法[J]. 机器人, 1995, 17(3): 129-139.
WANG Xiaogang, QIAN Wenhan. A NEW METHOD OF DECISION FUSION FOR DISTRIBUTED MULTISENSOR: DOUBLE-HEAD TEST[J]. ROBOT, 1995, 17(3): 129-139.
Citation: WANG Xiaogang, QIAN Wenhan. A NEW METHOD OF DECISION FUSION FOR DISTRIBUTED MULTISENSOR: DOUBLE-HEAD TEST[J]. ROBOT, 1995, 17(3): 129-139.

分布式多传感器决策融合的新方法──双端检验法

A NEW METHOD OF DECISION FUSION FOR DISTRIBUTED MULTISENSOR: DOUBLE-HEAD TEST

  • 摘要: 利用传统的Bayesian决策理论和假设检验方法处理分布决策融合问题有一个重要的缺陷就是缺乏柔性,无法将不确定与不知道区分开来;Dempster-Shafer证据理论虽然可以弥补这一缺陷,但作为其数学基础的公理化定义的严密性值得怀疑.因此需要有更完善、更可靠和更有效的统计决策及证据组合方法.在这方面Thomopoulos已提出了一种广义证据处理方法(GEP),它由Bayesian理论与D-S理论相结合发展而来.本文则在Neyman-Pearson检验方法基础上作了改进,提出了一种双端检验法(DHT),用于统计决策与证据组合.同GEP方法相比,本文的方法不需要待验假设的先验概率知识,故有更加广泛的适用性.

     

    Abstract: Using classical Bayesian decision theory and hypothesistest method with reference to problem ofdistributed decision fusion,an important limitation is lack of flexibility since the method can't distinguish be-tween uncertainty and unknowing. Though Dempster-Shafer theory may satisfy requirement of flexibility,but it lacks rigorousness in the axiomatic definition of evidence through independent experiments. Thus,more effective method is demanded.Thomopoulos had presented a new Generalized Evidence Processing(GEP)theory. He attempted to unify the Bayesian theory with the Dampster-Shafer theory.In this paper,we present a new Double-Head Test(DHT)method that was derived by improving NeymanPearson test. The DHT is more applicable than GEP because DHT needn't exact knowledge of the a -prior probabilities of the tested hypotheses.

     

/

返回文章
返回