Reserve policy optimization for scheduling wind energy and reserve

The rapid increase in the integration of renewable resources has given rise to challenges in power system operations. Due to the uncertainty and variability of renewable generation, additional reserves may be needed to maintain reliability. Uncertainty complicates the process of economic dispatch and renders the deterministic optimization approach less effective. Existing optimization solutions for handling uncertainty, such as scenario-based stochastic programming and robust programming, are also computationally expensive, especially when there are multiple wind farms. Such approaches are less practical for large-scale systems during real-time operations. This paper investigates offline stochastic algorithms to train deterministic operational policies. Such policies are then added to real-time operational models. Specifically, an offline policy generation technique is proposed to provide a stochastic reserve margin to hedge against the real-time uncertainty of (multiple) wind farm generation. The proposed policy generation structure uses a forecast-based framework that accounts for wind generation and system loading conditions. The proposed approach is tested on the Reliability Test System 1996. The proposed approach is compared against existing reserve rules to demonstrate the improvement in handling uncertainty and achieving a more secure solution.