MIMO System Identification for Large Power Systems Preserving Dominant Oscillatory Dynamics

This paper describes a general multi-input multi-output (MIMO) system identification approach for large scale realistic power systems. Low level probing is performed in time domain simulation to generate the input and output data, which is used to identify a low order MIMO model with a subspace method. The data can also be field measurements from low level probing on real power systems. The identified low order MIMO model preserves dominant oscillatory dynamics of the original system while taking the selection of inputs and outputs into consideration; thus it can potentially be used for designing a central oscillation damping controller. This paper illustrates the procedure and techniques to identify and validate the MIMO model. We apply the proposed approach to interarea oscillations but the framework can also be generalized for local oscillations. The approach is applied and validated on a 9-bus test system as well as on a 20,289-bus practical power system. The consistent dynamic performance between the original system and the identified low order MIMO model shows the effectiveness of the proposed approach.