PV FORECAST

A suite of AI machine-learning tools that accurately analyses and predicts solar irradiance.

PV Forecast is a suite of AI machine-learning tools, integrated into a sensor input system, that analyses and predicts solar irradiance in realtime. Energy foresight opens up revolutionary new ways to manage our energy systems.

Model on a range of reliable satellite sources and historic energy data

As the original primary source of data, we work with several different sources of satelite prediction (NOAA, NCEP, GFS) which guarantees the robustness and reliability of the base prediction.

AI constantly refines the prediction model through machine learning.

Once established, the AI intelligently refines that base model through a variety of numerical models cross-referencing it with data from locally placed sensors.

Enables optimal efficiency and more stable operations for PV systems

An accurate, realtime model of solar production days in advance gives a massive advantage to operators and homeowners — allowing them to plan and coordinate energy schedules, creating greater stability.

3 levels of foresight range and frequency.

Interactive Tab / Graph inserted about here, to demonstrate the differences in prediction range and the frequency of updates.

Historic Data

Life is busy, so it's a good thing Foodie takes less than 15 minutes per batch to manage. The rest is taken care of for you. Sweet automation.

Irradiance Sensor

Increase short-term prediction accuracy with data from locally installed irradiance sensors.

24 Hours Accuracy Improvement • 100km capacity

Prediction of PV irradiance/power output corrected in real time
Current data from the given day is taken and based on the immediate difference (measurement in the previous hour), the forecast for the rest of the day is corrected (with decreasing importance of the correction weight) (1h learning => correction for the rest of the day).
Increased precision of short term (until end of a day) prediction
Reacts on sudden weather changes

Sky Scanner

Track and predict cloud patterns up to 60 minutes in advance to optimise...

Short term • 10 second intervals • AI-powered motion tracking

Sky-imaging camera
Precise tracking of cloud movement
Cloud coverage processing in 10 seconds interval of calculation
Prediction is commuted using a special motion prediction model

Born and developed in Prague.

Built by the UCEEB in the Czech Republic.

SEE DEVELOPER

Take advantage of energy foresight.

Discover service

How much land do I need to build a 5 MW Solar farm?