An efficient prony-based solution procedure for tracking of power system voltage variations

Magnitude and duration are important characteristics when performing tracking of voltage variations. Accurate characterization of voltage variations relies much on the precise identification of frequencies of the measured signals. Although the Prony's method can provide high resolution for frequency estimation, the computational burden in the root-finding process is a crucial problem. A modified Prony-based solution procedure for voltage variation tracking is proposed in this paper. By providing a set of filters, the transfer polynomial with high estimation order in Prony's model can be efficiently reduced. The performance of the proposed method is validated by testing the generated and actual measured voltage signals. Results are compared with those obtained by typical Prony's method, namely, fast Fourier transform, adaptive linear neural network, and Kalman filtering. It shows that the proposed method is more accurate regardless of interferences of the power system frequency deviation, harmonics, and interharmonics, where both voltage variations and harmonics can be simultaneously detected.