Efficient robust nonlinear estimation technique for power quality indices measurement

This paper describes a new two stage robust Newton type numerical algorithm for power quality assessment in electric power systems. In order to estimate current and voltage spectra and the fundamental frequency simultaneously, in the first algorithm stage, the robust Newton type algorithm is applied. The algorithm robustness is achieved by developing an extra module called 'bad data detector'. In the second algorithm stage power quality indicators are calculated. Particularly the active, reactive, apparent and distortion powers are determined. As the main advantage, the technique provides estimates not sensitive to frequency deviations and bad data often appearing as a consequence of the communication error, incomplete measurement, errors in mathematical models etc. The algorithm performance is tested under laboratory conditions using the distorted voltage and current signals digitized during an AC-motor start.