AI-Driven Systems Slash Energy Use in Plant Factories, Boosting Sustainable Food Production

By Soham NandiReviewed by Joel ScanlonSep 11 2024

Cutting-edge AI optimizes lighting and climate control, achieving up to 32.34% energy savings while enhancing resource efficiency in diverse global climates.

Study: Artificial intelligence can regulate light and climate systems to reduce energy use in plant factories and support sustainable food production. Image Credit: Aleksandar Malivuk / Shutterstock

In an article published in the journal Nature Food, researchers explored how artificial intelligence (AI) and computational modeling optimize lighting and climate regulation in plant factories with artificial lighting (PFALs) across diverse global climates.

By reducing energy consumption, AI lowered energy use from 9.5–10.5 kilowatt hours per kilogram (kWh/kg) to 6.42–7.26 kWh/kg, depending on the climate. While AI improved energy efficiency, it also impacted other resource use, such as carbon dioxide (CO2), ultimately supporting more sustainable food production.

Background

The increasing challenges posed by population growth and urbanization have highlighted the need for resilient food production systems. One solution is PFALs, which offer year-round crop production in controlled environments. However, these systems consume significant energy and resources, and previous studies have shown conflicting results regarding their efficiency across different climates. This paper uniquely addresses these gaps by combining AI with advanced computational models to optimize resource use more effectively than previous studies, which often relied on static configurations and lacked comprehensive resource optimization.

This paper addressed these gaps by integrating AI and computational modeling to streamline environmental regulation in PFALs. It demonstrated that AI reduced energy consumption by 25% while maintaining optimal plant growth conditions.

Methodology for PFAL Optimization

The researchers detailed various components used to achieve optimal crop growth in PFAL systems. They first described a computational model based on Van Henten's two-state lettuce-growth model and validated energy and mass balance equations. The model assumed uniform environmental conditions within a PFAL constructed from a 40-foot shipping container, with light-emitting diode (LED) lighting, temperature, humidity, and CO2 controls.

The AI framework used deep reinforcement learning (DRL), specifically chosen for its superior ability to optimize resource usage like electricity while promoting lettuce growth. The DRL agent made decisions based on a Markov decision process (MDP) and considered external climate conditions, lighting schedules, and other operational factors. It adjusted key inputs like lighting, ventilation, and cooling to maximize crop growth while minimizing control costs.

The AI framework was compared against a baseline proportional control method, which followed pre-established grower rules. The study showed the DRL agent's potential to improve energy efficiency while ensuring optimal crop conditions, resulting in an average energy reduction of 32.34% across all tested climates.

Climate-Specific Outcomes

The researchers evaluated how AI reduced energy consumption in PFAL systems across ten cities with distinct climate conditions. PFALs were typically airtight, utilizing artificial lighting for photosynthesis and minimizing dependence on location.

However, ventilation was essential for providing oxygen during dark periods, affecting energy usage based on external climate conditions. The authors compared energy consumption using AI versus baseline strategies by assessing ten diverse cities, including edge cases like Reykjavik and Dubai.

AI consistently outperformed the baseline in energy efficiency, reducing energy use by an average of 32.34%. Cities with extreme temperatures, such as Miami, Phoenix, and Fargo, showed higher energy usage due to AI's adaptive ventilation, lighting, and cooling/heating systems handling. The AI system optimized energy by adjusting lighting intensity and ventilation, especially as plants matured. In cooler climates like Reykjavik, AI provided more significant energy savings, while hot, arid regions like Dubai saw less efficiency due to constant ventilation.

Additionally, AI helped manage CO2 utilization more effectively in milder climates, with warmer locations showing poorer CO2 efficiency. The researchers concluded that AI could substantially reduce energy costs, particularly in cooler regions, but cost savings were contingent on local climate conditions and energy prices.

Resource Efficiency and AI in PFAL Operations

The researchers highlighted the role of sustainable food production in achieving the UN's sustainable development goals (SDGs), particularly through improving resource efficiency in PFAL systems.

By leveraging AI and computational modeling, the authors explored optimizing energy use and ventilation strategies to conserve resources. Ventilation, a crucial yet often underestimated factor in PFAL operations, was found to be crucial for energy savings and plant health, reducing CO2 waste and enhancing oxygen supply during the dark period.

Additionally, AI-based optimization improved photosynthesis efficiency by mitigating photorespiration and regulating temperature. The results demonstrated that energy use in PFALs could be reduced by approximately 25%, offering significant environmental and economic benefits, especially in diverse climate conditions. However, the research acknowledged limitations, such as uniform indoor condition modeling, and suggested future refinements for larger facilities. The study also pointed out that integrating AI into real-world PFALs poses challenges, such as the need for advanced cyberphysical infrastructure and effective training methods like simulation-based training (sim2real) to ensure successful deployment.

Conclusion

In conclusion, the researchers demonstrated that AI and computational modeling could significantly enhance resource efficiency in PFAL operations, particularly by optimizing lighting and climate regulation. AI-driven systems reduced energy consumption by up to 25% depending on the climate while maintaining optimal plant growth conditions.

Additionally, the use of AI improved CO₂ utilization and ventilation efficiency, contributing to more sustainable food production. Although challenges remain, such as scaling up for larger facilities, the findings offer valuable insights into the potential of AI to improve energy efficiency and sustainability in controlled environment agriculture.

Overall, this research not only advances the understanding of AI's role in PFALs but also provides a pathway toward more sustainable, efficient, and resilient food production systems that can adapt to diverse global climates.