On-line System Identification of Power System Linear Models

This paper deals with the provision of adequate testing simulation environments to facilitate the evaluation of new system identification techniques to enhance power system stability. It develops and implements a testing architecture for online system identification approaches, taking advantage of the use of two well-known computational programs (Matla$b^{TM}$ and DIgSILENT PowerFactor$y^{TM}$). To ensure a reliable and optimal system identification, this platform exploits a mutually beneficial relationship among four well-established mathematical techniques (discrete Fourier transform, Teager-Kaiser energy operator, the fast Fourier transform, and eigensystem realization algorithm). Their symbioses can capture the synchrophasor information, detect the right disturbance instant, optimally extract the dominant frequency, and properly identify the Markov parameters that assemble the descriptor form of Multiple-Input and Multiple-Output (MIMO) linear systems for modeling modern power grids. Numerical results unveil the potential for recreating the real behavior of power systems; in particular, the proposed architecture can deal with any system identification technique to be tested.