Integrating wind variability to modelling wind-ramp events using a non-binary ramp function and deep learning models
This work addresses the challenge of managing wind variability for grid operators, potentially reducing reliance on fossil fuels and associated costs, though it appears incremental as it builds on existing machine learning methods.
The paper tackles the problem of forecasting wind power ramp events to integrate wind energy into electricity grids, proposing a novel approach that incorporates high-resolution wind fields and deep learning models to improve classification and prediction.
The forecasting of large ramps in wind power output known as ramp events is crucial for the incorporation of large volumes of wind energy into national electricity grids. Large variations in wind power supply must be compensated by ancillary energy sources which can include the use of fossil fuels. Improved prediction of wind power will help to reduce dependency on supplemental energy sources along with their associated costs and emissions. In this paper, we discuss limitations of current predictive practices and explore the use of Machine Learning methods to enhance wind ramp event classification and prediction. We additionally outline a design for a novel approach to wind ramp prediction, in which high-resolution wind fields are incorporated to the modelling of wind power.