Wind Energy Forecast in Complex Sites with a Hybrid Neural Network and CFD based Method

The wind is an intermittent renewable energy source and the energy production forecast is a fundamental activity for many reasons (grid regulation, maintenance, etc.).In this work a hybrid method (based on weather forecast data, neural networks and computational fluid dynamics) and a pure neural network approach are compared in a complex terrain site.The post processing of real production data has been discovered to be a key activity. Treatment and filtering of data spreading out from the supervisory control and data acquisition system are fundamental both for training and testing methods reliability.