DeepMind: Google uses AI technology to boost wind energy efficiency

The AI technology enables Google to predict the output of its central US wind farms more than a day in advance

Posting an update to on its company blog this week, Google’s staff confirmed that its decision to apply machine-learning algorithms across 700MW of wind power capacity last year had paid off.

The algorithms, developed thorugh DeepMind – the AI solutions firm owned by Google’s parent company Alphabet Inc – enable Google to predict the energy output of a wind farm 36 hours in advance, considering factors such as weather forecasts and energy market fluctuation.

Deepmind’s system consists of neural networks – computer systems modelled on the human brain. Google has trained these networks to predict the variability of wind power outputs and energy market price fluctuations. Once these predictions have been made, the technology recommends the actions which Google staff should take to make optimal hourly delivery commitments, ensuring both grid stability and that the company receives the highest possible price for its clean power.

“We can’t eliminate the variability of the wind, but our early results suggest that we can use machine learning to make wind power sufficiently more predictable and valuable,” DeepMind’s programme manager Sims Witherspoon said.

“Our hope is that this kind of machine learning approach can strengthen the business case for wind power and drive further adoption of carbon-free energy on electric grids worldwide.”

Google has been powering 100% of its operations with renewable power since April 2018. Last year, 99% of Alphabet Inc’s $111bn revenue was accounted for by clean energy and other low-carbon technologies.

Data centres

To date, Google has only applied the DeepMind technology to its renewables projects in the central US region, but the tech firm has expressed a desire to roll the programme out further in the coming months and years. Google is also continuing to use the DeepMind platform to directly control its data centre cooling systems as a way of lowering emissions and energy consumption.

In 2016, Google announced that using a combination of AI and employee training had enabled it to reduce energy consumption at its data centres by 40%. Since then, it has moved to let DeepMind implement its own recommended changes, rather than those made by staff, in a bid to drive further progress.

In a bid to help other corporates harness AI and machine learning to achieve their sustainability goals, Google last year launched a $25m (£19m) Impact Challenge, enabling organisations of all sizes to submit concepts for how AI can be used to alleviate and address key societal and environmental challenges.

Sarah George

Action inspires action. Stay ahead of the curve with sustainability and energy newsletters from edie