State-space system identification-toward MIMO models for modal analysis and optimization of bulk power systems

Abstract

This paper provides an introduction to a reduced-order, small-signal identification approach to modal analysis and control of large power systems. Being based on system-wide responses to low-energy pulse excitations generated using conventional time-domain simulation software such as PSS/E or EMTSP, it readily takes full advantage of the large built-in model database. The proposed multi-input-multi-output (M1MO) minimal realization reveals naturally the dominant modes attached specifically to a given device, as well as the transfer functions relating selected measurement and observation sites. It plays a complementary role to direct computation of the full-scale linearized model using a comprehensive program such as MASS. After a summary of the theoretical work initiated at Hydro-Québec in the early 1990's to promote this approach and put it into routine use, we present the main challenges in developing a production grade computer code. Detailed examples inspired by actual network studies at HydroQuebec are discussed, the most complex of them involving the identification of a 125th order MIMO model with 26 inputs and 26 outputs,. © 2000 IEEE.