A method for detecting non-stationary oscillations in process plants

• Oscillations are detected in non-stationary time series.
• Signal baseline computation procedure is introduced to handle non-stationarity.
• Robust statistics are used to address noise and data uncertainties.
• The method is tested with industrial measurement data from a board machine.

Persistent oscillations are a common problem in process plants since they cause excessive variation in process variables and may compromise the product quality. This paper proposes a method for detecting oscillations in non-stationary time series based on the statistical properties of zero-crossings. The main development presented is a technique to remove a non-stationary trend component from a signal before applying an oscillation detection procedure. The properties and performance of the method are analyzed using simulation experiments, a comparative study using industrial benchmark data, and tests with paperboard machine data. Finally, the simulation and industrial results are analyzed and discussed.