ISA transactions
-
Principal component analysis (PCA) is a popular fault detection technique. It has been widely used in process industries, especially in the chemical industry. In industrial applications, achieving a sensitive system capable of detecting incipient faults, which maintains the false alarm rate to a minimum, is a crucial issue. ⋯ A dynamic PCA model is also developed for boiler leak detection. This new method is applied to boiler water/steam leak detection with real data from Syncrude Canada's utility plant in Fort McMurray, Canada. Our results demonstrate that the proposed method can effectively reduce false alarm rate, provide effective and correct leak alarms, and give early warning to operators.