Synchrophasor estimation research has recently shifted emphasis from the phasor to the frequency and rate of change of frequency estimation problem. The main concern now is to achieve finer dynamic spectral analysis from the power system signals. This paper proposes the use of the Taylor-Fourier filters to reduce the error of the whole set of estimated parameters. This reduction is attained by exploiting their capability to match the phase pattern of the signal from one estimation to the next. By being endowed with their own phase follower, they are able to point their basis vectors toward the given signal, providing the best possible estimates from the nearest available subspace. The adaptive filters, provided with their own polynomial phaselocked loop, follow the instantaneous frequency of the given signal. Since most of the signals stipulated in the synchrophasor standard are defined by Taylor polynomials, the performance of this family of filters is remarkable, for durations from two cycles. They obtain estimates with zero error in the full set of parameters, when the test signal is in the focused Taylor-Fourier subspace. Very small estimation errors can be obtained for signal to noise ratios greater than 50 dB. Harmonic interference can also be constrained, but with higher computational cost.Index Terms-Dynamic phasor estimates, frequency estimates, phase-locked loop (PLL), phasor measurement unit, power system oscillations, ROCOF, synchrophasors.
0018-9456