The Impact of Meteorological Forcings on Agri-Photovoltaic Systems: Advances in Greenhouse Energy Modeling

Agri-photovoltaics (Agri-PV) represents a new strategy for maximising land use by integrating solar energy production into the agricultural operation. Greenhouses, which are essential to modern agriculture, provide a controlled environment that promotes plant growth, extends the growing season and protects crops from extreme weather conditions. The integration of photovoltaic (PV) systems into these greenhouse structures is a forward-thinking move toward sustainable and energy-efficient agricultural practises. The REGACE project, funded by Horizon Europe, aims to develop and validate a new technology, i.e. a responsive tracking system placed inside the greenhouse driven by a PLC controller that changes the angle of the tracking system according to the needs of the plants, while enabling the continuous production of renewable energy in greenhouses throughout the year, supporting uninterrupted food production without the constraints of energy availability.

In this progressive and dynamic context, we are advancing the development and testing of a Digital Twin (DT) ecosystem. This DT ecosystem simulates the energy production of PV greenhouse systems under specific internal conditions and external meteorological forcings. This paper presents our results from the use of Deep and Shallow Neural Networks (DSNN) to model the environmental conditions in these greenhouses. These networks improve our understanding of microclimatic variables and can lead to more general control strategies that optimise both crop yield and energy consumption.

Preliminary results indicate that Deep and Shallow Neural Networks (DSNN), like traditional complex physical models, can effectively model the internal conditions of greenhouses with relatively low error and strong correlationpredictive capability, but with less computational effort. These results were derived from data collected during the summer at one of the pilot sites in Greece. The DSNN skilfully captures the variability of internal thermodynamic parameters, which is crucial for the management of external environmental influences and internal systems such as cooling and shading.

The next steps aim to integrate the modelling of plant, water and energy balances within the greenhouse environment. This extended model will be further validated in the first pilot plant in Greece as well as at five other greenhouse sites involved in the project. This step is crucial for the development of integrated and adaptive control strategies that can improve both agricultural productivity and energy efficiency under different climatic conditions.