An Optimized Subspace-Based Approach to Synchrophasor Estimation

We present two improvements to the subspace-based phasor measurement unit (PMU) algorithms based on ESPRIT frequency estimation. The first is a dynamic, real-time thresholding method to determine the size of the signal subspace. This allows for accurate ESPRIT-based frequency estimates of the nominal system frequency and the frequencies of any out-of-band interference or harmonic frequencies. Since other frequencies are included in the least squares (LS) estimate, the interference from frequencies other than nominal can be excluded. This results in a near flat estimation error over changes in: 1) nominal system frequency; 2) harmonic distortion; and 3) out-of-band interference. We also reduce the computational burden of ESPRIT and demonstrate the proposed algorithm running in real time on resource-constrained platforms. The accuracy of this algorithm is evaluated using the IEEE standard for synchrophasor estimation, IEC/IEEE 60255-118-1, and compared with another state-of-the-art PMU algorithm.