On steady-state model uncertainty in phasor estimation: A robust statistical approach

The poor performance of the DFT phasor estimator in off-nominal frequency scenarios is very well-known in the synchrophasor estimation literature. Similar effects are observed due to harmonics, interharmonics, and negative-sequence unbalances in three-phase signals. It is shown that all these issues are a direct consequence of a model mismatch between an underlying statistical model and the actual signal. In order to cope with this model uncertainty, a robust phasor estimator (RPE) is proposed. It is obtained by minimizing the worst-case mean squared total vector error (MSTVE), a novel metric which generalizes the TVE to account not only for the estimator bias due to the model mismatch but also for the estimator variance due to noise. Numerical results show that a significant improvement in performance can be obtained with respect to the classical DFT and WDFT phasor estimators. Finally, an RPE design example is given, illustrating the flexibility and advantages of the robust statistical approach.