Model Predictive Control via Output Feedback Neural Network for Improved Multi-Window Greenhouse Ventilation Control.

Area of Science:

• Agricultural Engineering
• Artificial Intelligence in Agriculture
• Horticultural Science

Background:

• Greenhouse environmental control is vital for plant growth, with temperature management being critical, especially in warm climates.
• Rapid temperature increases in greenhouses necessitate effective ventilation strategies.
• Optimizing multi-window ventilation is complex due to diverse window structures and independent operation.

Purpose of the Study:

• To develop and evaluate a novel ventilation control logic for multi-window greenhouses using an output feedback neural-network (OFNN).
• To enhance temperature regulation accuracy and optimize window-opening behavior for strawberry production environments.

Main Methods:

• Development of a prediction model utilizing an output feedback neural-network (OFNN) with 15 input variables.
• Implementation of an optimization algorithm based on OFNN to manage six window-opening behaviors.
• Validation through three case studies and simulations on a nonlinear model.
• Field experiment comparing the developed control logic against conventional methods in a six-day trial.

Main Results:

• The OFNN prediction model achieved high accuracy with an R-squared value of 0.94.
• Optimization algorithm demonstrated effective control over window-opening behaviors.
• Field experiment showed a reduction in Root Mean Square Error (RMSE) from 3.01 °C (conventional) to 2.45 °C (OFNN).

Conclusions:

• The OFNN-based ventilation control logic significantly improves temperature management in multi-window greenhouses.
• The developed system offers superior performance compared to conventional ventilation control strategies.
• This approach provides a promising solution for precise environmental control in horticultural production.