W. FENG, C. Ruiz Mora, A. Alcántara Mata
Wind energy plays a crucial role in the energy mix, with its integration requiring optimal wind farm siting and turbine sizing to enhance output, mitigate fluctuations, and improve system stability. This study examines two wind farm location portfolios, leveraging a Quantile Neural Network (QNN) to estimate the aggregated power output distribution. The QNN is embedded as a constraint in an optimization framework to determine optimal siting and sizing. To address wind power uncertainty while balancing energy revenue and investment costs, two stochastic optimization models are proposed: a risk-averse model using Conditional Value at Risk (CVaR) to mitigate worst-case losses and a risk-seeking model using Conditional Value at Best (CVaB) to capture high-profit scenarios. A case study in Asturias, Spain, demonstrates that optimal wind farm configurations vary with risk preferences, offering insights into strategic decision-making for wind energy investments under uncertainty.
Palabras clave: Wind power quantiles, Constraint learning, statistical risk measure, Data-driven optimization
Programado
Optimization and Learning in Energy
11 de junio de 2025 15:30
Auditorio 2. Leandre Cristòfol