Forecasting renewable energy in Italy: comparative approaches for energy planning

The energy transition represents a key challenge for the 21st century, requiring robust forecasting methods to facilitate the integration of renewable energy sources into national power systems. Italy, with its rapid deployment of solar photovoltaic and wind power, presents an ideal case study for forecasting methodologies to inform policy, infrastructure development, and investment decisions. This thesis investigates statistical and machine learning models for forecasting the installed capacity of renewable energy in Italy, focusing on solar PV and wind power. The research, conducted between 2022 and 2025, evaluates di˙erent methodological approaches, compares their performance, and discusses implications for national and regional energy planning. Chapter 1 introduces the global and European framework of the energy transition, with particular attention to the EU Green Deal, long-term scenarios (TYNDP 2024), and the alignment with Italian national strategies. Chapter 2 examines the historical evolution and current status of renewable energy deployment in Italy, highlighting trends in PV and wind capacity, regional disparities, and challenges for grid integration. Chapter 3 presents the forecasting methodologies, including regression-based approaches, Grey models, and machine learning techniques such as Feedforward Neural Networks and Long Short-Term Memory. Comparative criteria and evaluation metrics are also discussed. Chapter 4 illustrates the application of these models to national and regional datasets (2006–2023), highlighting the performance of di˙erent approaches and pre-senting case studies on wind (Tuscany–Apulia) and solar PV (Emilia-Romagna–Apulia). Conclusions are discussed in Chapter 5.