Intra-day and Day-ahead Wind Farm Output Forecasting using Neural Network Ensembles

摘要

Wind energy is an increasingly important component of a utility's service offerings. Due to the intermittent nature of wind energy, the accuracy of wind farm output forecasts is critical to ensuring optimal integration of wind energy with other sources on a grid. GDF SUEZ has developed an innovative approach to improving the accuracy of wind farm output forecasts which involves developing ensembles of neural networks, each of which is tuned to the characteristics of its target area. An overview of neural network technology and the neural network ensemble modeling process is provided, along with preliminary results based on actual operating data from GDF SUEZ in Lyon France.