PCA-SDG在TEP多源故障诊断中的应用

Application of PCA-SDG Based Multiple Fault Diagnosis

针对传统基于SDG(符号有向图)的故障诊断方法对每个变量节点状态和高低阈值难以确定,且对各个变量单独统计,不考虑变量间相互关系的缺点,提出一种PCA(主元分析)与SDG相结合的故障诊断方法,并将其用于多源故障诊断中。将PCA得到的出现故障征兆的变量在SDG模型上进行反向推理,找到故障源。通过TEP仿真实验验证,表明该方法能够及时有效地检测出单个或多个故障,提高了诊断的准确性与分辨率。

The fault diagnosis using SDG （signed directed graph） is uncertain about node state and threshold value ,and performs single variable analysis without considering correlation of var/ables. A method combining PCA（principle component analysis） and SDG was proposed to improve the traditional multiple fault diagnosis. We can single out variables with the failure symptom from PCA, backward inference on SDG model and find the possible fault root（s）. The TEP case studies show that the method can find one or more faults fast. imoroving the accuracy and resolution.