Short-term forecasting and predictive control of rooftop greenhouse microclimate using multi-horizon machine learning models

Urban Farm DB summary

A data-driven framework for rooftop smart greenhouses was developed, combining environmental data, meteorological inputs, and actuator states for short-term temperature forecasting. Seven machine learning models were compared, with the best performing model achieving R² > 0.98 and MAE between 0.280 and 0.311. A fuzzy logic-based predictive control system demonstrated over 60% energy savings while maintaining climate stability, though challenges for generalization include dataset size and geographical variability.