Modeling of environmental parameters in greenhouse with Linear Time-Invariant System theory

In order to reduce the massive consumption of water and energy on greenhouse environmental control in summer. The environmental model is needed for precisely control of temperature and humidity. In the paper an Auto-Regression (ARX) model of indoor temperature and humidity under forced ventilation in summer was established by utilization of system identification technology. The model has 8 input parameters: the indoor/outdoor temperature and humidity, the outdoor solar radiation intensity, the outdoors instantaneous wind rate, the state of the forced ventilation system(on or off) and time readings within a 24-hour cycle. All data was acquired with one minute interval from June to July in 2003. The data was divided into two sets: one is identification set, which was used to identify the models, another is confirmation set which was used to verify the models. Model parameters were identified by least square method. The model was verified by the indices of maximum absolute error(MAE), maximum relative error(MRE), RMSE and variance accounted for (vaf). The confirmation showed that the MAE of the temperature models was 3.57℃, RMSE was less than 0.198℃; the MAE of the humidity models was 7.3%, RMSE was less than 0.624%; the vaf of models was up to 98.9%. The Models have higher precision in general, whereas the predicted error of the models is large a little bit at several samples, and can satisfy the demand of greenhouse environmental control.