Análise do desempenho de novos métodos de predição das componentes de Clarke aplicados na estimação digital de frequência em sistemas elétricos de potência

Abstract

The electrical frequency is a parameter of great importance for the full functioning of the Electric Power Systems (EPS), influencing the functioning and operation of the equipment and the quality of the energy supplied. This work presents innovative methods for digital frequency estimation in EPS. The estimation techniques are based on the analysis of the network voltage waveforms, which are decomposed into their α and β components using the Clarke Transform. The future values of the α and β components are predicted through three methodologies: Derivative Functions, Polynomial Interpolation Based with Newtonian Basis and Linear Prediction. From these values, the frequency of the network is then estimated as a function of the angle resulting from the product between the Clarke complex signal and that given by the prediction of the α and β components. The proposed methods were tested for frequency signals with ramp, exponential and damped sinusoidal variations. The methodologies were evaluated in terms of convergence time and minimum and maximum errors before and after convergence, showing that the proposed techniques have great precision and robustness in the face of simulated situations. Key-words: Digital Frequency Estimation. Protection of Electrical Power Systems. Clarke’s Transform. Derivative Functions. Polynomial Interpolation with Newtonian Basis. Linear Prediction. Performance Indices.