Wind power forecasting

Forecasting energy production from wind farms is a classic problem of the renewable energy industry. Electricity producers need to know the power output in advance to support optimal system management and energy trading activities.

However, unlike conventional electricity sources, wind resources can vary significantly over the course of the day and seasonally. Hence, accurate forecasting is important for keeping wind energy relevant and profitable.

In this project, we set out to build artificial intelligence (AI) powered models for forecasting wind power production on an hourly basis across six wind farms.

Wind energy produced over a given period of time depends heavily on the wind speed, air density, and other meteorological variables as well as the energy conversion efficiency of the turbines. We employ a data-driven approach to forecasting hourly wind power involving detailed feature engineering, modeling, and interpretation of results.

The power forecast for one of six wind farms is displayed below and a summary of project results can be found here.