MIMO Identification of Power System With Low Level Probing Tests: Applicability Comparison of Subspace Methods

Dynamic modeling of a power system is fundamental to its safe operation and control. With the development of wide-area measurement system (WAMS) and low level probing tests, it is feasible to estimate a multiple-input multiple-output (MIMO) model of a power system without large disturbances. In practice, outputs during the low level probing tests consist of measurement noise, system noise and probing responses. Before applying a field experiment of MIMO modeling, noise influence to the system modeling as well as applicability of existed identification methods should be examined. In this paper, four kinds of subspace methods are carefully compared for their applicability in power systems' MIMO modeling with low level probing test. Results with different noise conditions explore the noise influence and the suitable methods for field experiment. Monte Carlo method is used for statistic analysis in a small system and conclusions are validated in a large complex system, China Southern Grid.