Root cause diagnosis of plant-wide oscillations using Granger causality

• The paper presents a new data-driven time series method for diagnosing the sources of plant-wide oscillations.
• The proposed method combines the Granger causality, spectral Granger causality, and principal component feature selection.
• Simulation and industrial case studies demonstrate that the effectiveness of the proposed method.

Oscillations are common in closed-loop controlled processes which, once generated, can propagate along process flows and feedback paths of the whole plant. It is important to detect and diagnose such oscillations to maintain high control performance. This paper presents a new data-driven time series method for diagnosing the sources and propagation paths of plant-wide oscillations. The proposed method first uses a latent variable method to select features which carry significant common oscillations, then applies both time-domain Granger causality and spectral Granger causality to provide reliable diagnosis of oscillation sources and propagations. Simulation tests and an industrial case study are shown to demonstrate the effectiveness of the proposed method.