No abstract

An on-line measurement of equivalent parameters of a distribution system and a harmonic generating load is discussed in the paper. Four parameters are the subjects of the measurement: the distribution system and the load complex impedances, the complex rms (crms) values J, of the load generated current j harmonics and the crms values E, of the distribution system internal voltage e harmonics.The method discussed is based on measurements performed in two states of the system, before and after disturbing it by a capacitor or by a harmonic generating device. The effects of the random and systematic measurement errors on the calculation accuracy are discussed in the paper. The results are illustrated with computer modeling and with physical measurements. U,= U,ejan, I,, = I, ejpn(1) denote complex rms (crms) values of the n-order voltage and current harmonics U, and i , at the load terminals in a circuit, as 0-7803-2615-6/95/$4.00 0.1995 IEEE

A method is discussed for on-line measurements of parameters of a poweir distribution system and its associated load. Four parameters are measured: the complex impedances of the distribution system and the load for harmonic frequencies; the complex rms (crms) vralues .L of the harmonics of the load current j and the crms values of the harmonics of the distribution system voltage e. The measurements are performed on two states of the system before and after it is disturbed by either connecting to it a capacitor or a harmonic generating device. The results of physical measurements are compared with those calculated from a model.

No abstract

The presentedpaper illustrates the invasive method of power-system identification. It is based on Czarnecki's ideas to disturb slightly the system and observe its response to find unknown harmonic sources and equivalent linear impedances. The one-phase system is considered, with a new signal-processing technique in which instant measurements are based only on two adjacent signal periods. That approach allowed to move differential calculations to the time domain, and to deal with integral Fourier transform, in place of discrete, in frequency domain. The new algorithm allows to increase the S/N ratio, is insensitive to signal time-shifts and gives correct results in a wide frequency band (5 kHz). Simulation and rneasurements results are presented and show how accurate the identification can be f o r short-time (two periods) and 1 s long observations.