动态辨识框架下条件证据更新的故障检测方法

在分布式传感器网络中,各个子网往往具有不同的辨识框架,此时经典的证据理论无法处理。针对这一问题,提出一种动态辨识框下的证据融合理论和条件更新理论的故障检测方法。首先获取最新的观测证据,提出采用模糊隶属度函数作为信任转换的桥梁,完成动态辨识框架下的信任测度;然后利用新来证据的信任测度对已有的证据进行更新,以此进行各个观测区域的故障检测;最后通过构造两个传感器子网S1和S2的分布式检测与识别系统对所提方法进行验证,结果显示该方法在处理动态辨识框架和故障检测方面的有效性。...

Full description

Saved in:
Bibliographic Details
Published in计算机应用研究 Vol. 32; no. 8; pp. 2370 - 2373
Main Author 吴祎 周强 吴文军 吴迪 胡胜
Format Journal Article
LanguageChinese
Published 陕西科技大学电气与信息工程学院,西安,710021%陕西延长石油安源化工有限公司,陕西榆林,719319%西安理工大学理学院,西安,710054%西安交通大学机械工程学院,西安,710049 2015
Subjects
动态辨识框架
故障检测
条件概率更新
置信测度
条件概率更新
动态辨识框架
dynamic FoD
fault detection
置信测度
conditional probability update
mass measure
故障检测
Online AccessGet full text
ISSN1001-3695
DOI10.3969/j.issn.1001-3695.2015.08.031

Cover

More Information
Summary:在分布式传感器网络中,各个子网往往具有不同的辨识框架,此时经典的证据理论无法处理。针对这一问题,提出一种动态辨识框下的证据融合理论和条件更新理论的故障检测方法。首先获取最新的观测证据,提出采用模糊隶属度函数作为信任转换的桥梁,完成动态辨识框架下的信任测度;然后利用新来证据的信任测度对已有的证据进行更新,以此进行各个观测区域的故障检测;最后通过构造两个传感器子网S1和S2的分布式检测与识别系统对所提方法进行验证,结果显示该方法在处理动态辨识框架和故障检测方面的有效性。
Bibliography:51-1196/TP
In the distributed sensor networks, individual sensors always have different FoDs, this lead the classical evidence theory can not be used. In view of this problem,this paper proposed a fault detection algorithm in the dynamic FoD using conditional probabilistic approach. Firstly, it derived the new observed evidence and its belief measure. Secondly, it used the conditional probability of new evidence to update the old belief measure. Thus, it could use the new updated belief measure to detect fault. Finally,it constructed a two-sensors-subnets model to verify this approach. Simulation results demonstrate that the proposed method shows an effective ability in dealing with the dynamic FoD and fault detection.
Wu Yi , Zhou Qiang , Wu Wenjun, Wu Di, Hu Sheng （1. School of Electrical ＆ Information Engineering, Shaanxi University of Science ＆ Technology, Xi＇ an 710021, China; 2. Shaanxi Yanchang Petroleum Mining Industry Co. , Ltd, Yulin Shaanxi 719319, China; 3. School of Sciences, Xi＇ an University of T
ISSN:1001-3695
DOI:10.3969/j.issn.1001-3695.2015.08.031