Presentations
Coherent probabilistic day-ahead forecasting of wind power generation
Max Bruninx, PhD researcher, Vrije Universiteit Brussel
Abstract
We introduce a novel, distribution-free framework for probabilistic day-ahead wind farm power forecasting. By extending the Quantile Regression Network with Incremental Quantile Functions and Inductive Conformal Prediction, we address common pitfalls in quantile regression. Our approach yields coherent quantiles and more reliable confidence intervals, enabling effective risk management and informed decision-making for wind farm operators.
