Data-Driven estimation of frequency response from ambient synchrophasor measurements

With the wide deployment of synchrophasor technology, measurement-based dynamic modeling and studies have been becoming increasingly useful for real-time grid operations. This paper considers the problem of estimating the power system frequency response from ambient synchrophasor measurements. Specifically, we develop the analytical conditions for establishing the equivalence between the cross correlation of ambient generator speed data and the system frequency response between any two locations. Our conditions, relying on uniformly damped and equally excited oscillation modes, extend earlier work on electromechanical wave propagation modeling to nonhomogeneous power networks. Numerical results suggest that the validity of the cross correlation approach would hold for more realistic conditions such as nonuniform damping and high-order generator model. Its practical value is further corroborated by real-data results, which closely match with the actual propagation time of electromechanical waves recorded during the 2008 Florida blackout in the Eastern Interconnection system.