METHOD FOR RECONSTRUCTING IMPULSIVE VOLTAGE DISTURBANCES BASED ON COMPETING HARMONIC APPROXIMATIONS

Abstract. This paper proposes a method for reconstructing impulsive voltage disturbances in electric power networks based on competing harmonic approximations. The approach relies on constructing multiple parametric signal models within a sliding time window and selecting the optimal model according to a minimum residual (approximation error) criterion. To enhance the robustness of parameter estimation, window functions with improved spectral leakage suppression properties are employed, along with iterative procedures for refining the estimates. Unlike traditional time–frequency methods (DFT, wavelet transforms, S-transform), the proposed scheme enables direct reconstruction of the time-domain waveform of an impulsive disturbance and allows quantitative estimation of its amplitude, duration, and time of occurrence. The algorithmic structure is presented, and the rules for forming the set of competing approximations as well as the criteria for their comparison are described. It is shown that the method is suitable for power quality monitoring applications and can be implemented in near real-time, provided that the number of harmonic components is limited and computational optimizations are applied.