Improving the efficiency of PQ analysis concerning quasistationary waveforms through special computational techniques

Computational techniques conceived to improve the efficiency of PQ analysis of quasi-stationary signals acquired from the secondary winding of the excitation transformer in a power plant are addressed. Signals of 30 periods (SN) were analyzed. SN were polluted by white Gaussian noise and "arithmetic average" signals (AS) of one period length were computed for 2 reasons: (a) simpler noise removal; (b) evaluation of mean values for the PQ indices corresponding to all periods involved in the computing of AS, saving runtimes. Auxiliary minimal data structures were conceived and used in order to deal with isolated spikes and notches. The periods containing them are submitted to "smoothing" procedures. The recalculated AS was used as reference to evaluate the periods with energetic deviations which have to be denoised individually in order to evaluate separately their PQ indices. Comparisons (classic vs original algorithms for PQ analysis) were made revealing the advantages and drawbacks.