The deepening penetration of renewable power generation is challenging how the minute balancing of supply and demand is carried out by power system operators. Several proposals to short-term operational planning rely on robust optimization to offer guarantees on the ability of the operator to meet a wide array of possible scenarios. The main downside of these approaches is their conservative results whose operating costs and/or carbon footprint may be sub-economical. Such results come by because these approaches immunize their solutions for the required level of security against realizations of potential events within their uncertainty set. Moreover, these approaches also often ignore the inherent time and spatial couplings of wind and solar generation variability. In this paper, we seek to reduce the conservativeness of the robust solution by proposing the concept of spatio-temporal flexibility requirement envelopes. We show how it is able to efficiently capture and model the temporal trends and spatial correlation of multisite renewable generation and load. A mathematical program for energy scheduling is also developed using the projections of this envelope. We showcase the use and advantages of spatio-temporal flexibility requirement envelopes and their associated scheduling approach in a microgrid and on a modified IEEE Reliability Test System.
With the deepening penetration of wind generation in power systems, it is urgent that wind turbine generators (WTG) can have active power control (APC) capabilities, i.e., WTG can adjust active power output according to the power command from wind farms. The improvement of APC performance is closely related to the rotor speed variation (RSV) mode and the optimal setting of stable equilibrium points (SEP). Compared with the active RSV with the same SEP, the passive RSV no longer focuses on the rotor speed tracking and has lower power response discrepancy and drive-train loads. However, limited by the realization principle, APC based on passive RSV overlooks the optimization and setting of SEP, which makes the rotor speed of WTG easily reach the bound of the variable-speed range and leads to speed overshoot and power drops. Hence, this paper proposes an APC method with additional pitch angle compensation. Consequently, WTG under passive RSV has the ability of optimizing and setting SEP. On this basis, the compensation pitch angle is dynamically optimized according to the predicted wind speed. The experimental results show that the proposed strategy can effectively reduce speed overshoot and power drops while taking full advantages of passive RSV.
This article presents a novel methodology to maximize wind farm power generation by integrating graph neural networks (GNN), supervised learning, and reinforcement learning techniques. First, the article introduces a graph‐based representation of the wind farm, capturing wind turbines as vertices and the inter‐turbine wake interactions as edges. The construction of this graph representation integrates the Jensen wake model, which includes insights derived from prior knowledge of wind farm aerodynamics. Subsequently, a detailed description of the GNN model's architecture, incorporating a message passing mechanism, is outlined. This GNN model is trained initially with supervised learning using a dataset of optimal pitch angles generated from the analytical results derived from Jensen wake model. Moreover, to improve the GNN model's accuracy and adaptability, reinforcement learning techniques are employed. The GNN model interacts with a high‐fidelity wind farm simulation environment, receiving feedback in the form of rewards derived from the wind farm's actual power output. Through a policy gradient approach, the GNN parameters undergo iterative updates, enabling the model to learn and adapt to dynamic wind conditions and intricate turbine interactions. The effectiveness and advantages of the proposed methodology are demonstrated through comprehensive case studies across various wind farm layouts.