Case Study

Predicting wind energy output

Case study: Predicting wind energy yield

We predicted the wind energy yield of a farm in Australia for a whole year with 80% accuracy.

Thanks to our methodology, the collaborative production of different energy sources can be coordinated more efficiently, so shortages or costly overproduction can be avoided.

The challenge

Wind farms are a key source of renewable energy. Timely and reliable prediction of wind energy output is critical for evaluating investment decisions for building new farms and for energy load balancing when coordinating production of traditional power plants and weather-dependent sites.

The increasing availability of weather data and more accurate weather forecasts in recent years motivated us to look for patterns and dependencies between weather features and wind energy outputs.

The goal we set
  • How can we estimate wind energy yield based on weather forecast data?

What we did?

We analyzed several months of data from an Australian wind farm together with the publicly available weather data measured at the same location. Robust predictive models helped us iteratively identify the MINIMAL LIST OF WEATHER FEATURES impacting the energy output.

Initially 16 features were used to predict energy output, out of which ONLY TWO FEATURES (the wind gust and the dew point) were identified as key drivers. We also quantified each of these factors’ contribution to the accuracy of predictions.

We developed predictive models using the data from October 2010 to June 2011, but the predictions were tested on data collected in July 2011, which is an entirely unseen season. The results showed a very good prediction accuracy (of 12% on rooted mean squared error) and allowed to forecast yields for the entire year from weather data, identifying worst and best-case scenarios.

"...The presented framework is so simple that it can be used literally by everybody for predicting wind energy production on a smaller scale—for individual wind mills on private farms or urban buildings, or small wind farms.”

The impact we created

With a reliable way to forecast wind energy output, energy suppliers can now coordinate the collaborative production of different energy sources more efficiently and avoid costly overproduction.

Related links/Publications:

Ekaterina Vladislavleva, Tobias Friedrich, Frank Neumann, Markus Wagner: Renewable Energy Journal, 2013, vol 50, p.236-243 Predicting the energy output of wind farms based on weather data: Important variables and their correlation.

Talk to us about how you can turn your data into a system to deliver success

Our core expertise is in business-driven applications of predictive analytics and data science to solve complex business challenges which directly impact the bottom line.