Design and application of multifrequency signals for power plant identification

The unpredictable behaviour of system signals and weak stationarity of plant dynamics are the most common problems in the online identification of power system components. Both these problems may be greatly reduced by the insertion of a suitable perturbation signal into an accessible point of the system under study to increase statistical activity. Advantages include improved stability of parametric estimates plus a reduction of computational overhead and a simplification of measurement procedure. The properties of perturbation signals for system identification are presented and discussed. A novel pseudo random signal which is based on properties of the Gaussian model is introduced and compared with the performance of some presently available perturbation signals.<>