SLIDE: Satellite Images Prediction with Deep Learning
We improve short term solar irradiation forecasts in the 15 minutes to 2 hours range, which is critical for off grid sites or isolated power grids, using AI and DL techniques.
What is the challenge that is being addressed?
Photovoltaic power is a growing source of energy. Although it is more and more used, its high penetration can challenge existing power grids as the power generation is weather-dependent. Especially, in low-connected systems such as off-grid industrial sites or island power grids, ramps of photovoltaic power are especially difficult to deal with.
What is the AI solution the project has implemented?
In order to help overcome the above challenges, we implement a model that provides regional forecasts of irradiance by forecasting the upcoming frames of satellite-derived irradiance map videos. We develop a model for Satellite Images prediction using a combination of state-of-the-art AI techniques that is designed to forecast irradiance maps. It uses up to four satellite-derived irradiance maps in order to generate forecasts of upcoming irradiance in a roughly 200x200 km area up to two hours ahead.
Who helped implement the AI solution?
This solution is implemented in the context of SR4C3, a winning project from the AI4Copernicus 1st Open Call. The consortium consists of SOLAÏS and Transvalor, while the results of the project are available here.