Dynamic state prediction based on Auto-Regressive (AR) model using PMU data

Fenghua Gao, James S. Thorp, Anamitra Pal, Shibin Gao

Research output: Chapter in Book/Report/Conference proceedingConference contribution

30 Scopus citations

Abstract

This paper presents a dynamic state prediction method based on an Auto-Regressive Model (AR model) using PMU data. In recent years, state prediction has played a key role in improving power system performance and reliability. When load is increased linearly at a constant power factor, it is proved in this paper that the bus voltages are quadratic and the AR model for predicting the next voltage is based on three prior estimates. This logic is then tested on the IEEE-118 bus system. The test results demonstrate that under morning load pick-up, economic dispatch, line opening and generator oscillations, the proposed method is correct and gives valid predictions. Furthermore, based on the error in quadratic fit, it is advocated that this method could be applied to detect abnormal conditions in the transmission systems. Theoretical analysis and results show that the proposed method based on AR model has great potential in predicting power system states.

Original languageEnglish (US)
Title of host publication2012 IEEE Power and Energy Conference at Illinois, PECI 2012
DOIs
StatePublished - May 17 2012
Externally publishedYes
Event2012 IEEE Power and Energy Conference at Illinois, PECI 2012 - Champaign, IL, United States
Duration: Feb 24 2012Feb 25 2012

Publication series

Name2012 IEEE Power and Energy Conference at Illinois, PECI 2012

Other

Other2012 IEEE Power and Energy Conference at Illinois, PECI 2012
CountryUnited States
CityChampaign, IL
Period2/24/122/25/12

Keywords

  • Auto-Regressive (AR) Model
  • Dynamic State Prediction
  • Phasor Measurement Units (PMUs)
  • State Estimation

ASJC Scopus subject areas

  • Energy Engineering and Power Technology
  • Fuel Technology

Fingerprint Dive into the research topics of 'Dynamic state prediction based on Auto-Regressive (AR) model using PMU data'. Together they form a unique fingerprint.

Cite this