The increasing availability of high fidelity, geographically dispersed measurements in power systems improves the ability of researchers and engineers to study dynamic behaviors in the grid. One such behavior that is garnering increased attention is the presence of forced oscillations. Power system engineers are interested in forced oscillations because they are often symptomatic of the malfunction or misoperation of equipment. Though the resulting oscillation is not always large in amplitude, the root cause may be serious. In this report, multi-channel forced oscillation detection methods are developed. These methods leverage previously developed detection approaches based on the periodogram and spectral-coherence. Making use of geographically distributed channels of data is shown to improved detection performance and shorten the delay before an oscillation can be detected in the online environment. Results from simulated and measured power system data are presented.
Revised: October 6, 2015 |
Published: September 30, 2015