Integrating Machine Learning for Crop and Energy Optimization in Agrivoltaic Gardening Systems

An Agrivoltaic gardening system offers a promising pathway to simultaneously addressing food security and renewable energy production through the integration of photovoltaic panels and crop cultivation. However, balancing crop productivity and energy generation remains a complex challenge due to variable microclimatic conditions and crop-specific responses to partial shading. This study presents a machine learning–based framework for optimizing both crop yield and photovoltaic energy output in small-scale agrivoltaics gardening systems. Using multi-source data including solar irradiance, soil moisture, temperature, panel configuration parameters, and crop growth indicators, supervised learning models were developed to predict crop yield and energy performance. Random Forest and Artificial Neural Network models demonstrated strong predictive capability, achieving coefficients of determination (R²) above 0.85 for crop yield estimation under shaded conditions. Multi-objective optimization revealed design configurations that improved combined land-use efficiency by up to 10% compared to conventional layouts. The results highlight the potential of machine learning to support adaptive design and real-time decision-making in agrivoltaics gardening, particularly in resource-constrained environments. This work contributes to emerging research on AI-enabled agrivoltaics and provides a scalable framework for sustainable food–energy co-production.