Ponente
Descripción
SUNRISE (Scalable Ultra-fast Nowcasting with Real-time In-Situ Embedded hardware) is a research project aimed at predicting photovoltaic solar energy output in real time at sub-minute timescales. The system combines all-sky camera imagery and irradiance sensor data with deep learning models (ConvLSTM and transformer architectures) to anticipate cloud-induced irradiance transients. A key objective is the deployment of these compressed AI models on low-power reconfigurable hardware (FPGAs/SoCs) directly at the edge, meeting strict latency and resource constraints. The project holds the European Commission's Seal of Excellence (ERC-2025-POC, ref. 101247189 – CHRONOS) and has industrial backing from IGNIS and TheNextPangea. This talk will present the project motivation, the modelling strategy, and the current status of the embedded prototype.