Researchers at Karlstad University have developed a new intelligent control strategy for battery storage in climate controlled greenhouses. By combining artificial intelligence and signal processing algorithms with short term forecasts of electricity consumption and solar power generation, energy costs can be reduced while simultaneously easing the load on the power grid.
Greenhouses provide optimal conditions for year‑round food production but are also highly energy‑intensive. In the present study, the researchers show how a battery storage system controlled by reinforcement learning can be used more efficiently when the greenhouse has a high share of locally generated renewable energy, such as solar photovoltaics.
“The results show that the optimised energy‑use system could be applied on a larger scale, for example in the manufacturing industry, where it is needed to reduce both production costs and carbon footprints,” says Jorge Solis, Associate Professor of Electrical Engineering at Karlstad University.
The system was tested using real operational data from a climate‑controlled greenhouse at Karlstad University, equipped with solar panels and battery storage.
Clear economic benefits
The results show that the adaptive control strategy outperforms both traditional fixed‑rule approaches and simpler AI models:
- Variable electricity costs were reduced by 2.2% in February and 2.7% in March.
- Peak power demand, which forms the basis for so‑called high‑load charges, was reduced by up to 24%.
- Total electricity costs were lower than in all the scenarios used for comparison.
Contributing to a more sustainable energy system
The research represents an important step towards more sustainable and cost‑effective food production, as well as more flexible energy systems that can better accommodate large shares of renewable electricity.
The study was carried out with support from the Swedish Energy Agency within the programme Energy Efficiency in Lighting.
Further research and new collaborations
“It is very encouraging that other industry stakeholders have already shown interest in further developing the proposed adaptive control of energy storage for large‑scale applications,” says Jorge Solis.
Read the study
