Abstract
Oscillations travel along propagation paths and may impact the control performance of the whole plant. This paper presents a data-driven method for diagnosing the root cause of the plant-wide oscillation.The major contribution is the application of combining Granger causality, a statistical method based on linear prediction theory and principal component feature selection to provide a reliable diagnosis of oscillation propagation. Two case studies are used to demonstrate our proposed method. The work presented here may have significant implication for diagnosing of other kind of disturbance propagation. © 2012 IFAC.
| Original language | English |
|---|---|
| Pages (from-to) | 160-165 |
| Number of pages | 6 |
| Journal | IFAC Proceedings Volumes |
| Volume | 47 |
| Issue number | 15 |
| DOIs | |
| Publication status | Published - Jul 2012 |
| Externally published | Yes |
| Event | 8th International Symposium on Advanced Control of Chemical Processes, ADCHEM 2012 - , Singapore Duration: 10 Jul 2012 → 13 Jul 2012 |
Funding
The authors are grateful for the financial support from China Scholarship Council (CSC) and the Center for Interactive Smart Oilfield Technologies (CISOFT), and thanks to Eastman Chemical Company for providing the industrial data.
Keywords
- Diagnosis
- Granger causality
- Oscillation
- Propagation
- Spectral analysis