Title :
Incipient fault detection and isolation in a PWR plant using principal component analysis
Author :
Kaistha, N. ; Upadhyaya, B.R.
Author_Institution :
University of Tennessee
Abstract :
A method for the detection and isolation of incipient faults in field devices in industry using Principal Component Analysis (PCA) is presented. Nominal operation data typically lie on a low-dimension surface due to relationships imposed by the physics of the process and are modeled using PCA. Abnormal deviations from the surface lead to fault detection while isolation is a consequence of these deviations being in different directions for different faults. A steam generator in a pressurized water reactor (PWR) is used for demonstration.
Keywords :
Data analysis; Databases; Fault detection; Feedback control; Matrix decomposition; Monitoring; Personal communication networks; Physics; Principal component analysis; Sensor systems;
Conference_Titel :
American Control Conference, 2001. Proceedings of the 2001
Conference_Location :
Arlington, VA, USA
Print_ISBN :
0-7803-6495-3
DOI :
10.1109/ACC.2001.946059