Assessing Frequency Variability Using Long Term High Reporting Rate Measurements

Renewable's integration in power systems contributes to decreasing the available mechanical inertia., which further impacts the frequency variability in steady-state conditions of operation. In this paper two metrics are chosen to be applied to the frequency signal available with high reporting rates (up to 50 frame/s)., exhibiting random variations. The metrics are the coefficient of variation of root mean squared error CV(RMSE) and the coefficient of determination
$R^{2}$
and are applied on different measurement (200ms., 1s) over different (10s) time windows. The higher selectivity of CV(RMSE) compared to
$R^{2}$
is demonstrated for synthetic signals emulating steady-state and dynamic operation and on a set of 10 minutes recorded frequency variation in four nodes of the Romanian Power System during a severe event recorded in 2017. Also, a procedure to discriminate power system events using a flag applied to a new signal (deviation of frequency from a model adopted on each measurement window) is proposed.