Wind is a highly variable and intermittent source of energy, which means that integrating it into the power system can prove tricky. Therefore, when integrating wind energy into the power system, power companies will rely heavily on backups to manage against the sudden changes in wind energy. To deploy this juggling act efficiently requires accurate forecasts of when the wind levels will rise and fall. But, at present, current wind forecasts are highly inaccurate. New research from the Nevada Solar-Energy-Water-Environment Nexus Project aims to use a combination of Big Data analytics and statistical models to provide more accurate, and therefore valuable, estimates of wind power. Initially, researchers needed to teach the model what has happened historically. To do this, they "trained" the model on a massive data set from a large wind farm. Sensors on this wind farm recorded the wind speed and direction, and instruments also recorded the output of each of the 300 wind turbines every 10 minutes. This model uses the forecast of the wind at projected future intervals and the probability of this value occurring to determine how best to maximize the productivity of the system. Simulations using this model have been shown to improve the efficiency of the system.
Visit Website | Image credit: Cary Leppert/Shutterstock.com