Fast robust power system dynamic state estimation using model transformation

Xueyuan Wang, J. Zhao, Vladimir Terzija, Shaobu Wang

Research output: Contribution to journalArticlepeer-review

10 Citations (Scopus)


Accurate information about generator rotor speeds and angles plays an important role for power system transient stability online assessment and protection. To address this need, this paper proposes a fast and robust estimation approach based on the model transformation strategy. Thanks to this strategy, the original complex nonlinear model is transformed into a linear one without linerization, which makes the dynamical system observability analysis and the estimation problem significantly easier to solve. The proposed model transformation strategy is achieved by taking the measured generator active power as the input variable and the derived frequency and the rate of change of frequency measurements from the phasor measurement units (PMUs) as the output variables of the dynamical generator model. This allows us to estimate the generator rotor speeds and angles using only local PMU measurements and the swing equations, relaxing the need of a detailed generator model on which the existing dynamic state estimators are based. A robust Kalman filter is also developed to handle data quality problems as the frequency and rate of change of frequency measurements can be biased in presence of severe disturbance or communication issues. Comparison results carried out on the IEEE 39-bus system successfully validate the effectiveness and robustness of the proposed approach under various conditions.

Original languageEnglish
Article number105390
JournalInternational Journal of Electrical Power and Energy Systems
Publication statusPublished - Jan 2020
Externally publishedYes


  • Bad data
  • Dynamic state estimation
  • Kalman filter
  • Model reduction
  • Power system dynamics
  • Rotor speeds and angles
  • Synchrophasor measurements


Dive into the research topics of 'Fast robust power system dynamic state estimation using model transformation'. Together they form a unique fingerprint.

Cite this