DETECTION AND DIAGNOSIS OF UNIT-WIDE OSCILLATIONS

This article addresses the detection and diagnosis of oscillations in process measurements. It uses the power spectra of the measurements because the spectra are insensitive to the phase lags caused by time delays and process dynamics. Principal components analysis (PCA) can be applied to the power spectra for detection of groups of oscillatory process measurements. In cases where a well defined periodic oscillation is detected then diagnosis of the root cause is possible using a measure of harmonic distortion. Examples using this measure are discussed and its suitability for diagnosis assessed.