Predictability of frequency excursions in fluctuation-driven power grids

In the energy transition, modern power grids are increasingly exposed to fluctuating inputs e.g. from renewable power generation. However, many aspects of the collective dynamical responses across power grids still remain unknown. Combining an analysis of the stochastic features of the injected noises and a theory of network dynamic response patterns, we study the predictability of the spatio-temporal responses in power grids driven by stochastic fluctuations. We find colored fluctuations with flatter spectra induce heterogeneous yet stationary power grid responses, while more strongly colored fluctuations induce homogeneous yet non-stationary responses. These findings help to evaluate the potential systemic risks in power grids induced by different classes of stochastic fluctuations.