论文部分内容阅读
针对传统基于PCA(主元分析)的传感器故障诊断方法缺乏故障推理能力,难以定位故障源的缺点,提出一种PCA与SDG(符号有向图)相结合的传感器故障诊断的方法。此方法分为2步:首先建立系统的SDG模型和PCA模型,使用PCA方法监控所有的过程变量;第二步,当故障发生的时候,通过PCA得到异常变量的状态,根据变量的状态,通过SDG模型进行反向推理,找到可能发生故障的传感器。通过液位控制系统的仿真实验以及在常减压装置(AUDU)故障诊断上的应用,结果表明方法能够及时有效地检测出单个或多个传感器故障,提高了诊断的准确性与分辨率。
In order to solve the shortcomings of traditional fault diagnosis based on PCA (Principal Component Analysis), a fault diagnosis method based on PCA and SDG (Symbol Directed Graph) is proposed. This method is divided into two steps: First, set up the system SDG model and PCA model, and use PCA method to monitor all the process variables; second, when the fault occurs, get the abnormal variable state through PCA, according to the state of the variable, The SDG model performs the reverse reasoning to find the sensor that is likely to fail. The simulation experiment of liquid level control system and the application of AUDU fault diagnosis show that the method can detect single or multiple sensor faults timely and effectively, and improve the diagnostic accuracy and resolution.