The paper investigates the performance of recently proposed frequency estimation methods in the presence of typical power system measurements disturbances, i.e. sinusoids and additive noise. The tests are conducted for the polyphase Prony method, the polyphase smart DFT, and the discrete-time frequency-gain transducer. Considered methods are capable of unbiased one-cycle frequency estimation. The robustness against disturbances is obtained by the application of the proposed prefilter. Consider methods with the initially filtered measurement signal perform significantly better than the standard PMU (phasor measurement unit) P class and M class signal processing models.
(Data obrony: 2008) Wetula, Andrzej Wydział Elektrotechniki, Automatyki, Informatyki i Elektroniki
Low-frequency disturbances are a separate kind of power system voltage disturbances. They should be interpreted as a complex modulation, or simultaneous amplitude and phase modulation. The thesis presents six measures of the low-frequency disturbances. These measures were designed to obtain high sensitivity, allow to build a quantitive scale, and be computationally effective in order to allow realtime measurements. Above requirements were met by most of proposed measures.Presented measures use a complex envelope signal, calculated using the Hilbert transform based algorithm. The algorithm was implemented in a digital signal processor and a PXI system, and tested using real power grid signals. Proposed algorithm was built for a low-frequency disturbances analysis. However, it is also useful in other kinds of analyses. As an example of another use of the algorithm, usage of calculated complex envelopes for an identification of Thevenin model of a power system was presented. Preliminary simulation results show usefulness of the algorithm in such a kind of analysis. Also, use of one of the developed measures for chosing credible identification results was presented.
