Machine learning for improving PV energy prediction to develop green and sustainable airports in Romania

Unofficial AI-generated summary based on the public title and abstract. Not an official translation.

🔬研究者:Compare ML models for PV forecasting; CNN outperforms other architectures in airport context.

🏢実務担当者:Adopt CNN-based PV forecasting to improve energy management and reduce operational costs at airports.

🏛政策担当者:Support policies that encourage ML-driven renewable energy integration in airport infrastructure.

Accurate photovoltaic (PV) energy forecasting is essential for ensuring reliable and sustainable power management, particularly in energy-intensive infrastructures such as airports. As airports face increasing energy demand and pressure to comply with European and national sustainability goals, integrating PV systems can substantially reduce both carbon emissions and operational costs. However, forecasting PV generation in airport environments remains challenging due to complex meteorological variability and the lack of localized predictive studies for Romania. This study applies machine learning techniques—including Deep Neural Networks (DNN), Standard Recurrent Neural Networks (SRNN), Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Convolutional Neural Network (CNN)—to forecast PV power generation using national-level data. After evaluating 20,401 unique model configurations, the CNN architecture achieved the best performance, with a coefficient of determination (R2) = 0.983 and a mean absolute error (MAE) of 13.7, outperforming GRU and the other models. The results confirm that advanced neural models can effectively capture both daily and seasonal patterns of solar generation, enabling optimized energy management for airport infrastructures. These findings provide a scientific basis for implementing intelligent energy systems that support Romania’s transition toward green and sustainable airports. Beyond its scientific contributions, this research has significant implications for both industry and society. The implementation of accurate PV forecasting systems based on advanced machine learning models can support airport authorities and energy operators in optimizing resource allocation, reducing operational costs, and ensuring more stable integration of renewable energy into the grid. At a broader level, such predictive frameworks contribute to accelerating the decarbonization of critical infrastructures and promoting environmental awareness within local communities. The methodology proposed in this study can also be extended to other large-scale facilities, fostering collaboration between research, industry, and policy-makers in achieving sustainable energy transitions.

• semanticscholar https://doi.org/10.1088/2631-8695/ae4db1first seen 2026-05-15 20:42:54