Starting point selection approach for power system model validation using event playback

Model validation is an essential task to determine whether a model can accurately describe the actual behaviours of a power system. Currently, major commercial software tools are equipped with an ‘event playback’ function to validate dynamic models using testing data from phasor measurement units (PMUs). Due to their limited bandwidth and low sampling rates, PMUs cannot capture the fast-transient dynamics. As such, the playback function using the conventional approach may mistakenly invalidate an accurate model during the high-frequency responses. To overcome the deficiency, a batch state estimation approach is proposed in this study to improve the performance of the ‘event playback’ function by focusing on low-frequency responses in model validation. The proposed approach consists of three major steps. First, a multi-model adaptive Kalman filter approach is used to estimate the dynamic states of the system. Second, the singular spectrum analysis (SSA) is used to detect the fault clearance time. Finally, the estimated states after the fault are used as the initial states of the ‘event playback’ to validate the dynamic model during the low-frequency responses. The analytical basis of the proposed method is also provided by showing the existence and uniqueness of the trajectory of the underlying model. The effectiveness of the proposed approach is demonstrated using the PSS/E.