When AI becomes a 'farm manager': How can agrophotovoltaics achieve a 'double harvest' of food and electricity?

Agrophotovoltaics (APV), as an innovative model integrating agricultural production with solar photovoltaic power generation, is an important pathway to address the dual pressures of food security and low-carbon energy. As the global energy transition accelerates and agricultural modernization advances, its advantages of “dual land use and food–energy co-production” have become increasingly prominent. It can achieve efficient land use while accommodating clean electricity output and crop growth, aligning with the needs of sustainable development. However, for a long time, traditional models have been constrained by bottlenecks such as imbalanced light distribution, lagging temperature and humidity regulation, inefficient manual operation and maintenance, and data fragmentation, making it difficult to balance energy and agricultural benefits and seriously hindering large-scale promotion.

Mehrdad GHAMARI and colleagues from Edinburgh Napier University, UK, published a review paper titled “From data to harvest: artificial intelligence and machine learning technologies revolutionizing agrophotovoltaic systems” in Engineering Agriculture (DOI: 10.15302/J-FASE-2026675). The study systematically reveals that artificial intelligence (AI) and machine learning (ML) technologies can comprehensively optimize APV systems, providing key technical references for low-carbon smart agriculture.

Against the backdrop of global energy transition and agricultural modernization, the study reviews the diverse applications of solar energy technologies in agriculture and identifies the notable advantages of APV in improving land-use efficiency, enhancing climate resilience, and promoting biodiversity. At the same time, it precisely identifies the core shortcomings of traditional models: imbalanced light distribution and lagging temperature and humidity regulation make it difficult to balance food and energy benefits; low efficiency of manual monitoring and delayed fault troubleshooting constrain system stability; and prominent data fragmentation fails to support precision management, seriously hindering large-scale promotion.

Based on a systematic analysis of 129 relevant global studies, the research team constructed an intelligent APV framework integrating the Internet of Things (IoT) and AI, identifying four core application scenarios: forecasting and early warning, control and monitoring, optimization and design, and predictive maintenance. The study found that advanced algorithms such as neural networks, reinforcement learning, and convolutional neural network–long short-term memory (CNN-LSTM) networks perform excellently. The prediction accuracy for solar irradiance and crop yield reached 0.96; leafy vegetable yields in experimental fields in India and France increased by 10%–18%; fault detection accuracy exceeded 91%; and operation and maintenance costs decreased by 25%–40%, enabling integrated analysis and dynamic optimization of “food–energy–data”.

The study also objectively identifies current research gaps: lack of high-quality agricultural datasets, insufficient model interpretability, weak interdisciplinary collaboration, poor regional adaptability, and inadequate consideration of socioeconomic factors, pointing out short-, medium-, and long-term breakthrough directions for future research. It emphasizes the need to promote technology implementation and adaptation to different agricultural scenarios through interdisciplinary integration, lightweight model development, and data security governance.

The deep integration of AI and ML is moving agrophotovoltaics from “proof of concept” to “scalable solutions”. By improving land-use efficiency, reducing operation and maintenance costs, and enhancing system resilience, this technology provides a new path for sustainable agriculture for the 2.5 billion agricultural population worldwide and frees up valuable land space for the further expansion of renewable energy.