Detection and Diagnosis of Plant-Wide Oscillations using Granger Causality

Vitor O. Thomaz and Celso J. Munaro


Cause–effect analysis, fault diagnosis, spectral analysis


This paper addresses the problem of detecting multiple os- cillations and tracking down their root cause. The oscillations are detected by two methods based on spectral analysis, and the power related to the oscillations is used to identify the signals more likely to be the root cause. Since this method of finding the root cause is prone to fail when multiple frequencies are detected, Granger causality and sub- set selection methods are proposed. After a presentation of the methods, examples are provided using both simulated and industrial data. The clustering of signals containing the same frequency of oscillations followed by the methods to indicate the root cause is shown to be a good approach for rapid troubleshooting using only historical data.

Important Links:

Go Back