research highlights
Robust Continuous-Discrete Extended Kalman Filter for Estimating Machine States with Model Parameter Uncertainties
by Pengxiang Ren ( PhD Student ), Dr . Hanoch Lev-Ari ( levari @ coe . neu . edu ) and Dr . Ali Abur ( abur @ coe . neu . edu )
An accurate knowledge of rotating machine states , ( e . g ., generator states such as rotor speed , rotor angle and transient voltages , as well as exciter field voltage ), is important for dynamic state estimation in largescale power systems . The performance of dynamic state estimators , such as the ( extended ) Kalman Filter deteriorates in the presence of moderate model parameter uncertainties . In practice , parameters of rotating machines are known only with limited accuracy , and their values may vary with time . Several methods have been proposed in the literature to mitigate the effect of model uncertainties on overall performance , such as the H∞ approach , the set-valued estimation approach , and the guaranteed-cost paradigm . However , most of these methods are designed for state estimation in linear systems , whereas the ( two-axis ) generator model we consider is highly nonlinear .
A Robust Extended Kalman Filter ( REKF ) is developed for estimating machine states in the presence of model parameter uncertainties . Our robust filter is derived as a solution to a minimum-maximum ( worstcase ) optimization problem : it minimizes a squared residual norm over the entire range of ( bounded ) possible model parameter errors . The resulting REKF , which consists of modified time- and measurementupdate steps , outperforms the standard Extended Kalman Filter ( EKF ) in the presence of moderate parameters inaccuracies .
The following figures provide generator state estimation results in the presence of uncertainties in the damping constant ( D ) ( Figure 1 ) and the reactance ( Xd ) ( Figure 2 ), comparing the proposed REKF algorithm with the ordinary EKF .
Figure 1 : SE with uncertain Damping Constant
Figure 2 : SE with uncertain Reactance with transient
Mark your calendars . Our 2016 site visit is coming up in November at the Crown Plaza in downtown Knoxville , TN . We hope to see you then .
7