Probabilistic Robustness Verified Data-Driven Transient Security-Constrained Optimal Power Flow

Data-driven methods present a promising approach to extracting security boundaries in transient security-constrained optimal power flow (TSCOPF) for modern power systems. Nevertheless, with the increasing penetration of renewable energy sources (RES) and the wide deployment of information and communications technology (ICT), exact robustness evaluation of the employed data-driven models is essential to achieve the robust TSCOPF by ensuring that embedded data-driven security constraints can withstand variations, uncertainties, and adversarial conditions, which is an NP-complete problem. To this end, this paper proposes a novel methodology for enhancing the robustness and computational efficiency of power system operations through the development of a Probabilistic Robustness Verified Data-Driven Transient Security-Constrained Optimal Power Flow (PRV-TSCOPF) framework. In particular, a well-trained data-driven transient security assessment model is first verified and transformed into linear probabilistic robustness-verified security constraints with enhanced computational efficiency enabled by a novel three-phase approximation strategy. Furthermore, a stepwise scheme is developed to achieve robust dispatch, where adaptive sampling intervals are designed to reduce the error of the three-phase approximation and facilitate convergence. The effectiveness of the proposed three-phase approximation and PRV-TSCOPF is demonstrated through extensive case studies based on the IEEE 39-bus, IEEE 68-bus, and Iceland 189-bus systems, respectively.