Mining PMU data for real-time insights into grid power flow properties

Summary form only given. A synchronous power grid in steady state operation is a power flow solution engine: its inputs are the demanded and injected powers at load and generation buses, its outputs the sinusoidal (quasi-) steady state behavior of voltages at all buses in the network. With wide-spread deployment of phasor measurement units, grid operators are able to monitor the time varying, windowed Fourier component of these voltages about the 50 or 60 Hz fundamental. In traditional power flow studies, these voltage phasors would be estimated by solving the power flow equations, given power demands and injections. The premise of this work is that phasor measurements will allow real-time, “model free” monitoring of the properties of the power flow. On an fast time scale, load variation includes a broad-bandwidth stochastic component that excites the system over a wide range of input directions. In this framework, a widowed array of output vectors (i.e., the phasor measurements) carries information regarding the conditioning of the powerflow. The work here develops a “model free” computation of the largest singular value of a PMU measurement array, and shows it closely tracks the largest singular value of the power flow Jacobian inverse, as computed with full system model information.