Come rain or shine: using AI to power PVs in all weathers

We all know the British weather is unpredictable – one minute it’s blue skies, the next it’s a downpour. But what if we could outsmart the clouds?

That’s exactly what we’re doing with a new AI-powered forecasting system in the Netherlands. Developed with our colleagues at E.ON subsidiary Essent, this technology processes live data from 13 different weather providers, using machine learning to predict solar power generation with incredible accuracy.

By making real-time adjustments for unexpected weather changes, it helps us manage the grid more efficiently while cutting costs and reducing energy waste.

Cloud cover has always been the biggest challenge for solar power forecasting. Until now, sudden changes in weather have made it difficult to predict how much electricity our solar farms will generate – leading to costly imbalances and inefficiencies. But in the Netherlands, our AI-driven system, NowCasting, is changing that.

By combining real-time weather updates with historical patterns, NowCasting can predict solar power production in 15 minute intervals, up to two hours ahead. It uses key weather variables – humidity, wind direction, temperature, cloud coverage and radiation intensity – to generate hyper-accurate forecasts.

The power of AI: strength in numbers

Rather than relying on a single model, our approach uses an ensemble of multiple machine learning models. Think of it as a panel of experts rather than a lone voice – each model brings its strengths, compensating for any weaknesses in others. The result is a more reliable and precise forecast that allows us to adjust to changes in real time.

AI-powered predictions = smarter energy trading

Grid stability is a challenge in most countries, including both the Netherlands and the UK, and energy imbalances (supply not meeting demand, for example) can be costly. Providers must report how much electricity they’ll feed into the grid, but without precise forecasts, discrepancies can lead to expensive penalties.

Our AI system minimises these risks by identifying deviations early, allowing us to sell excess energy at the right time or purchase shortfalls before prices spike.

And it doesn’t stop there – the future of solar forecasting is set to get even smarter. We’re exploring the use of satellite data to track cloud formations in real time, adding an extra layer of accuracy for those unpredictable weather moments.

Smarter forecasts, greener future

With AI-powered predictions, we’re making solar energy more efficient, cost-effective, and reliable. That means a more stable grid, fewer wasted resources and a step closer to a net-zero future.

As solar power continues to expand in the UK, this cutting-edge forecasting technology could be just what we need to harness the power of the sun – without leaving it to chance.