基于PCA-SDG的水轮机调节系统故障诊断

Fault diagnosis of hydro turbine regulation system based on PCA-SDG

讨论了基于符号有向图SDG的故障诊断方法,建立了闭环控制系统的SDG模型及故障诊断推理规则.利用经验方法建立了典型的水轮机调节系统的SDG模型及故障诊断推理规则.针对SDG模型节点符号确定的主观性问题及节点阈值的组合爆炸问题,将主元分析PCA方法与SDG模型结合用于故障诊断.利用系统运行数据建立主元分析PCA模型,根据PCA模型构造残差统计量进行故障检测,当有故障发生时,将每一检测分量对残差统计量的贡献率与设定的贡献率阈值比较,得到各检测分量的定性符号值,该符号值作为SDG模型中相应节点的符号,根据SDG推理规则进行故障推理,找出故障源.根据现场运行数据建立水轮机调节系统的PCA模型,应用PCA-SDG故障诊断方法对其进行故障诊断,模拟系统传感器恒偏差故障,对故障诊断过程进行仿真,结果证明该PCA-SDG故障诊断方法对水轮机调节系统是有效的.

The method of fault diagnosis based on signed directed graph(SDG) was presented.The SDG model and inference rules for closed loop control system were established.For hydro turbine regulating process,the model of fault diagnosis and inference rules based on SDG was established by an experience approach.The principal component analysis(PCA) method combined with SDG model was applied to solve the subjectivity problem in the process for determining the node symbols and the ' combination explosion' problem of the node threshold.Firstly,PCA model was established based on system operating data,then the residual error statistics were constructed for fault detection.When fault occurs,contribution rate of each detection component on residual error statistics will be compared with the contribution rate of setting threshold value to obtain the qualitative symbols value of each detection component.The symbol value can be assigned to the corresponding node in the SDG model.Then,fault inference can be conducted according to the SDG inference rules to find out fault source.This PCA-SDG based fault diagnosis method was applied to analyze sensors diagnosis of hydro turbine regulating process,constant deviation fault of sensor was introduced to simulate,and the results prove that this method is practically feasible.