Amidst centralized and distributed frequency control in power systems

We propose a novel frequency control approach in between centralized and distributed architectures, that is a generalized continuous-time feedback control version of the dual decomposition optimization method. Specifically, a convex combination of the frequency measurements is centrally aggregated and followed by an integral control, which is broadcast as the control signal, and then by the optimal local allocations. We show that our controller comprises many previously proposed strategies for specific parameter sets. Under mild parametric assumptions, we prove local asymptotic stability of the closed-loop equilibria of the power system, which is modeled as a nonlinear, differentialalgebraic, dynamical system that includes traditional generators, frequency-responsive as well as passive loads, where the sources are already equipped with primary droop control. Our feedback control is designed such that the closed-loop equilibria of the power system solve the optimal economic dispatch problem.