Data-Driven Situation Awareness for Power System Frequency Dynamics

As the penetration of renewable energy increases, system inertia decreases, causing changes in system frequency dynamics. The power industry desires situation awareness of power system frequency dynamics to ensure secure and economic operation and control. Moreover, FNET/Grideye has abundant measured data from power systems, making it possible to conduct data-driven situation awareness studies on power system frequency dynamics. This doctoral dissertation proposes several contributions: (a) Two accurate generator trip event MW estimation methods are proposed, in which one is based on long window RoCoF and another is based on multi-Beta values; (b) Two real-time system inertia estimation approaches are developed using ambient frequency fluctuation and pump turn-off events, along with techniques for improving RoCoF calculation in event-based inertia estimation; (c) An adaptive PV reserve estimation algorithm is established to provide PV reserve while saving energy for PV resources; (d) A practical load composition estimation tool is built for the industry to easily obtain essential load model parameters. Although conducting research using actual data from power systems for practical application is challenging and compilated, the proposed data-driven situation awareness methods in this doctoral dissertation solve practical problems and offer clear theoretical explanations for the industry. These methods address one of the key challenges for operating a high-renewable power grid and pave the way for the U.S. carbon-free power sector by 2035.