Best Paper Award at IEEE PES APPEEC 2025 for the Department of Energy
Professor Sonia Leva and Professor Emanuele Ogliari, faculty members of the Department of Energy at Politecnico di Milano, together with PhD candidate Binh Nam Nguyen, received the 1st Place Best Paper Award at the international IEEE PES APPEEC 2025 conference (Asia-Pacific Power and Energy Engineering Conference), held from 2 to 5 December 2025 at the University of Auckland (New Zealand).
The award was granted by the organizing committee based on the scientific quality of the paper, the presentation, and the discussion of the results, and represents an important recognition of the Department’s research activities in the fields of solar energy, artificial intelligence, and advanced forecasting systems.
The awarded paper, entitled “A Multi-Source Deep Learning Framework for Very Short-Term Solar Irradiance Nowcasting with Satellite Data and All-Sky Images,” proposes an innovative approach to very short-term (15-minute) solar irradiance nowcasting, a key element for the efficient management of photovoltaic generation and the power grid.
A multi-source framework for solar nowcasting
The paper introduces a multi-source deep learning framework that integrates information from:
satellite data on cloud coverage and cloud type,
locally acquired high-resolution all-sky images,
supporting meteorological measurements.
The model is based on a Convolutional Neural Network (CNN) architecture with dedicated branches for the different input modalities, designed to synergistically exploit spatial information at the regional scale and local cloud dynamics. The system was validated on 17 months of data collected at two measurement sites, including the Bovisa campus of Politecnico di Milano.
The results show a significant improvement over the persistence model, the standard benchmark for very short-term forecasting, with a forecast skill of up to +27% under clear-sky conditions and robust performance also in more complex meteorological scenarios.











