Data-driven distributed optimal control of VPP for guaranteed frequency response in low-inertia power systems

In this study, we propose a data-driven distributed optimal control approach to realize guaranteed system frequency response (SFR) from aggregated resources within the virtual power plant (VPP). First, a framework for guaranteed SFR based on the concept of VPP is proposed, which includes an upper optimization layer and a lower coordinated control layer. The optimization layer provides each VPP with the frequency response target expected by the transmission system operator (TSO). Then, in the coordinated control layer, DERs can be controlled to realize the frequency response target. Second, the problem is formulated as a robust output containment control problem of heterogeneous leader–follower multi-agent systems with disturbances. Finally, a data-driven distributed robust containment controller is proposed for DERs. An off-policy integral reinforcement learning (IRL) algorithm is developed to obtain the optimal controller, which does not require any knowledge of system dynamics.