When dealing with renewable tech intermittent power supply is a problem for the national grid.
Being able to predict the factors that affect renewable energy – like the wind – is going to be critical as more renewable energy is used. Without huge energy storage infrastructure, it is tricky to plan how much power wind turbines can provide.
Alphabet – Google’s parent company – acquired London based artificial-intelligence company DeepMind back in 2014. At the time the company was known for developing complex artificial-intelligence programs to solve complex problems.
In a new blog post Goolge has laid out how DeepMind latest project has looked into better predicting wind power.
While existing AI and non-AI modelling techniques have existed for years it is still difficult to predict wind output more than a few hours ahead.
DeepMind and Google set out to be able to predict wind output 36 hours in advance and it looks like they have succeeded.
The results from DeepMind application of machine learning to Google’s wind power is impressively accurate
DeepMind trained its new algorithm on historical weather data alongside wind power output recorded over a year by Google own wind turbines.
The company believes its new model should help wind-farm owners make hourly commitments to the regional power grid manager at least a full day in advance.
Google believes that this latest project not only demonstrates the value of AI but “strengthen[s] the business case for wind power and drive further adoption of carbon-free energy on electric grids worldwide”.
