A hybrid data-model driven approach for time-decoupled power flexibility aggregation of heterogeneous distributed energy resources

Power flexibility aggregation provides an efficient way to harness the significant flexibility of large-scale distributed energy resources (DERs). However, the inherent heterogeneity and temporal coupling of DERs make it challenging to directly aggregate power flexibility, leading to conservatism and computational inefficiency. Therefore, we propose a novel hybrid data-model driven approach to characterize the aggregated active and reactive power flexibility region. The design of the proposed aggregation model consists of two stages. In the first stage, the active power related constraints of DERs are represented as high-dimensional polytopes. These polytopes are then inner-approximated by hyperrectangles to decouple the temporal coupling constraints. In the second stage, leveraging the mathematical models of DERs, we develop a computationally efficient data selection approach to obtain high-quality aggregated power boundary samples. The convex hull of these feasible samples is subsequently constructed to accurately characterize the aggregated active and reactive power flexibility region. The numerical simulations demonstrate that the proposed aggregation method significantly improves both the accuracy and the computational efficiency of the approximate flexibility region.