Estimation of uncertainty bounds for the future performance of a power plant

Peter Fogh Odgaard, Jakob Stoustrup

Research output: Contribution to journalJournal articleResearchpeer-review

13 Citations (Scopus)
283 Downloads (Pure)

Abstract

Prediction of the future performance of large-scale power plants can be very relevant for the operators of these plants, as the predictions can indicate possible problems or failures due to current operating conditions and/or future possible operating conditions. {A problem in predicting} the future performance of these plants is that available models of the plants are uncertain. In this paper three schemes for predicting uncertain dynamical systems are presented. The schemes estimate upper and lower bounds on the system performance. Two of the schemes are statistically based, one only based on recent data and the other is based on operating points as well. The third proposed scheme uses dynamical models of the prediction uncertainties, like in H-infinity-control. The proposed schemes are subsequently applied to experimental data from a coal-fired power plant. {Two sets of data from an actual power plant} are used, one containing normal plant operation and in the second set, coal is accumulating in the coal mill due to {an unbalance in the operating conditions.} These tests showed that {Scheme II} and {Scheme III} did bound the real
 system performance, while {Scheme I} failed doing so. In addition, the plant was simulated operating under the same conditions with additional large disturbances. These simulations were used to investigate
 the robustness and conservatism of the proposed schemes. In this test {Scheme I and II} failed, while {Scheme III} succeeded the test.
Original languageEnglish
JournalIEEE Transactions on Control Systems Technology
Volume17
Issue number1
Pages (from-to)199 - 206
ISSN1063-6536
DOIs
Publication statusPublished - 2009

    Fingerprint

Keywords

  • Dynamic Prediction
  • Uncertain Dynamical Systems
  • Power Plants
  • Statistical Prediction
  • Uncertainty Models

Cite this