Evaluation of Forced Oscillation Detection Methods

Forced oscillations can cause large power oscillations in power systems, it is therefore important to be able to accurately detect them on time. This paper evaluates the accuracy of frequency and damping ratio estimation from three methods for detecting forced oscillations. The Stochastic Subspace Identification method, a spectral method using Welch periodograms, and the self-coherence method. Simulations are performed on a small test network with full and partial, respectively, PMU observability when a forced oscillation is implemented as a sinusoidal load. The results indicate that all three methods are able to detect a forced oscillation with good accuracy, even when it has the same frequency as an inter-area mode. It was also found that the average estimation error increases somewhat when there is only partial PMU observability, and that selection of PMU locations can have an impact on the estimations.