Abstract
This paper investigates the use of data-based model validation for detecting the source of forced oscillations in power systems due to power plant governors. The underlying idea is that models that correctly reproduce the behavior of a component under normal conditions will not explain the behavior if the component is causing forced oscillations. During normal operation, a parametric method is applied in order to identify the transfer function between a generator’s electrical power and angular rotor speed. During a forced oscillation, we analyse the residual between the measured angular speed of the rotor and the simulated angular speed of all generators in order to locate the source of the forced oscillations. This method is comparable to a similar non-parametric methods. Moreover, the use of frequency measurements instead of rotor speed measurements is also investigated. The methods are tested and compared on the Kundur two-area system, with different locations, amplitudes and frequencies of the forced oscillation. The simulations show good performance for the parametric method when using angular speed measurements, and slightly poorer performance when using frequency measurements.