摘要：AbstractOscillations occurring in industrial process plants are often indicators of severe disturbances affecting the process operation. An accurate detection of oscillations is therefore of great interest for the economic viability of the process operation. The use of standard established oscillation detection methods requires however that the analysed signals fulfil conditions that are not always met in practice. A typical example is the presence of transient changes superposed to the oscillation pattern. This paper proposes simple heuristic methods to effectively detect and remove two types of commonly encountered transient changes (step changes and spike changes) from oscillatory signals. The effectiveness of the approach is analysed and the subsequent performance improvements of a standard oscillation detection method: the auto-correlation function method (Thornhill et al., 2003) is demonstrated. The approach is carried out on a 1,3-Butadiene production process where several measurements showed an established oscillation occurring after a production level change.
