Back to home
Condition monitoring of sensors with PCA method in nuclear power plants LI Wei1, PENG Minjun1, LIU Yongkuo1, and MA Zhanguo1 1.    College of Nuclear Science and Technology, Harbin Engineering University, 150001, Harbin, Heilongjiang, China (success870323@126.com, heupmj@163.com, lyk08@126.com,mazhanguo2013@163.com) Abstract: With the widespread application of digital I&C systems in Nuclear Power Plants (NPPs), more sensors are used to obtain operating information. A principal component analysis (PCA) method is applied in this paper to carry out condition monitoring for sensors in a NPP. Meanwhile to improve the model performance, a false alarm reducing method is proposed which is combined with PCA method in this paper. Sensor measurements from a real NPP are used to train and test the PCA model. Simulation results under normal operating condition indicate that the proposed false alarm reducing method really makes contribution to the model performance. Meanwhile, artificial failures with different degrees are sequentially imposed to test the functionality of the proposed PCA model, and the simulation results show that the proposed PCA model which is combined with a false alarm reducing method is effective on the condition monitoring of sensors no matter with major or small failures. Keyword: NPP; sensor failures; PCA; condition monitoring; false alarm reducing
	Click here for viewing full text.