Google has announced it has found a way to boost wind energy efficiency by a fifth with artificial intelligence (AI) forecasts.
The technology giant says it has successfully used its DeepMind machine learning platform to predict wind power output 36 hours in advance, reducing the variability and intermittency of the renewable power source.
It has trained a neural network using weather forecasts and historical turbine data, making wind farms more predictable and therefore valuable to the energy grid.
It uses the data to give wind farm operators more data-driven assessments of how to meet upcoming electricity demand and recommends how to make optimal hourly delivery commitments to the power grid a full day in advance.
Google has recently announced plans to build a solar project suspended above fishing ponds in Taiwan and is buying electricity from 1.6 million solar panels to power two of its new data centres in the US.