基于扩展证据理论的信息融合方法在传感器故障诊断中的应用

电厂传感器的网络结构提供了在空间或时间上的冗余或互补的信息，为传感器故障的检测和分离提供依据。证据理论的组合运算是在同一个识别框架下进行的，而电厂的传感器是来自不同层次上的信息源，其识别框架是不同的。为此提出一种在不同识别框架下的证据组合规则，采用精细和粗化两种算予，可以充分利用来自不同层次传感器的所有信息，减小传感器状态判断的模糊性。仿真结果表明，该方法可以对传感器的多故障进行有效准确的诊断。图2表2参6

Application of the Expanded Evidence Theory in Fault Diagnosis of Sensors by Using the Information Fusion Method

The network structure of power plant sensors provides complementary information, redundant in space or time, which may be used for detecting faults in sensors and isolating them. Combined calculations with the evidence theory proceed in the same framework of discrimination, but power plant sensors are at different hierarchic information levels, bound to different frameworks of discrimination. Therefore a rule for combining evidences that belong to different frameworks of discrimination is being proposed, which, with the help of both refining and coarsening operators, may abundantly utilize all the information coming from sensors belonging to different hierarchical levels, thereby reducing the judgment fuzziness concerning the state of the sensors. Simulation results show that this method is capable of accurately diagnosing multiple faults happening to sensors. Figs 2, tables 2 and refs 6.