Deep Blue is an R&D company operating on a European scale, focused on the role of the human in Safety Critical and high-tech systems particularly in the transport sector, and has recently entered the energy domain to develop innovative and sustainable solutions for the production and distribution of renewable energy. In this context, the REGREEN pilot aims to create a software solution to support the EU transition away from fossil fuels and towards carbon neutrality by developing an AI-based solution to identify the best energy generation sites of three renewable energy sources: solar, wind and hydropower. The objective is to produce a map of the sites which are suitable for an efficient production of at least two out of three sources. 

Challenge

Availability, intermittency, and unpredictability hinder the penetration of renewables. Wind and solar, the two most common sources, have more fluctuating production patterns than those derived from conventional power generation sources as their availability depends on weather. This poses a challenge at two levels: availability and stability. Electric energy storage is the common solution to the problem of energy unavailability, e.g. during the night or when wind is not blowing, but this solution shifts the problem of depletion of resources from fossil fuels to the materials needed to build batteries. The alternative option is the direct distribution of solar or wind energy to a grid, but this requires a system that is able to always balance the generated energy with the demand. 

Work Plan

REGREEN proposes a third and innovative alternative: using multiple-energy-source production as a means to enhance reliability and transmission stability. In a production site with at least two sources of energy, the probability of all being unavailable is low and the system can be optimised to compensate for the fluctuation locally and guarantee a stable energy transmission to the grid. REGREEN combines satellite observations, reanalysis, ground elevation and data about renewable energy power output and feeds it to an artificial intelligence algorithm that learns to identify the most promising sites for solar, wind and hydro energy production. The purpose is to produce a map of the energy generation sites where at least two out of three sources can be effectively produced. This will be achieved in steps. The first is to collect the data and identify the best performing AI model (December to April). The second is to train, test and validate the model (May to July). Subsequently, a map of the results is to be produced and displayed through an ad hoc visualisation interface (August to September).