Scientists have discovered a way to peek inside the "black box" of artificial intelligence to make wind power forecasts more reliable, according to Tech Xplore.
Wind power has one big challenge: predicting how much electricity turbines will generate daily. When those predictions are off, grid operators scramble to fill the gap with more expensive, often dirty fuel-based energy.
However, a team at Switzerland's École Polytechnique Fédérale de Lausannes fixing this problem with explainable artificial intelligence. The researchers trained a neural network using weather data and information from wind farms across Switzerland and worldwide.
|
Should we be harnessing the ocean to power our homes? Click your choice to see results and speak your mind. |
Absolutely 👍 "We tailored four XAI techniques and developed metrics for determining whether a technique's interpretation of the data is reliable," says Wenlong Liao, the study's lead author and a postdoc at EPFL's Wind Engineering and Renewable Energy Laboratory.
This approach helps power companies trust their wind forecasts by showing how those predictions are made.
"Power system operators won't feel very comfortable relying on wind power if they don't understand the internal mechanisms that their forecasting models are based on," explained Jiannong Fang, an EPFL scientist who co-authored the study.
Stay warm all winter long with the Apple of intelligent space heaters Kelvin is the Apple of space heaters, designed for energy efficiency and maximum comfort. It's completely silent and intelligently controlled, with setup taking just 5 minutes. Built from premium materials like aluminum and glass, Kelvin works beautifully as a full-home heating system or as the perfect solution for that one room that never feels warm enough. |
This means lower energy costs and cleaner air for everyday people as grids can reliably use more wind power. Cities and power companies can plan better and avoid using backup fossil fuel plants, saving money and reducing air pollution that affects our health.
While this AI approach helps make clean energy more affordable and reliable, it's worth noting that AI itself has environmental costs. Data centers housing these systems use substantial electricity and water for cooling: about 2 liters of water per kilowatt-hour of energy used. Training just one large AI model can consume enough electricity to power 120 average U.S. homes for a year, reported MIT News.
The team's findings were published in Applied Energy, and the XAI approach is already being implemented at wind farms to improve daily energy forecasts. As these systems spread, renewable energy sources like wind will become even more dependable parts of our energy mix, making cleaner air and a cooler planet achievable goals.
Join our free newsletter for weekly updates on the latest innovations improving our lives and shaping our future, and don't miss this cool list of easy ways to help yourself while helping the planet.
TCD Picks » Stasher Spotlight
💡Stasher's reusable food storage options make it easy and affordable to live life with less plastic
Cool Picks
