Energy: Photovoltaic, AI based method to forecast production

Increasingly accurate predictions of photovoltaic energy production thanks to artificial intelligence (AI) is the outcome of a study published by an ENEA research team in the scientific journal Energies.

“We have demonstrated the efficacy of our approach using actual production data from a photovoltaic plant located at the ENEA Research Center in Portici (Naples), which showed improved accuracy of photovoltaic production forecasts ^[1]”,

explained study co-author Amedeo Buonanno, researcher in the ENEA Smart Grid and Energy Networks Laboratory at the Department of Energy Technologies and Renewable Sources. “This,” he said, ”is essential for maintaining power grid stability, optimizing the operation of generation, load and storage resources and reducing operating costs.

To achieve this result, metereological forecast models that estimate photovoltaic power generation have been combined with a machine learning algorithm that integrates historical generation data, thereby increasing the accuracy of the forecast. This new ENEA method to forecast PV output offers significant advantages in terms of both versatility and applicability. Its generality ^[2] allows its use in a wide range of scenarios, even with reduced data availability as in the case of new installations.

 “The approach we propose is suitable for solar systems of different sizes, including small-scale systems installed in apartment buildings. For the latter, an initial prediction model can be developed based on the technical characteristics of the system (such as power rating and panel orientation), then refined through machine learning techniques and the use of historical generation data. Once trained, the models that yield the best results require relatively limited computational resources to generate accurate predictions. This feature facilitates their implementation, greatly expanding the possibilities for practical application in different operational contexts,” Buonanno pointed out.

Italy has made significant progress in the renewable energy sector, with installed capacity of photovoltaic systems exceeding 30 GW in 2023, marking a 21 percent increase over 2022 ^[3]. However, solar radiation variability is still a major challenge in managing PV power generation.

“The ENEA study is part of the MISSION^[4] project, which aims to develop innovative integrated energy systems by optimizing the interaction between different energy sources (renewable and conventional) and energy carriers (electric and thermal). Through smart and coordinated management, based on demand analysis and production forecasts, the project pursues to maximize the overall efficiency and improve the operational management of energy microgrids, one of the most promising models of electricity system transformation, thus accelerating the transition to a more sustainable energy future,” concluded Maria Valenti, head of the ENEA Smart Grid and Energy Networks Laboratory and MISSION project contact person.