Prediction of temperature and relative humidity in greenhouse based on the combination of DGM(1, 1) and linear model

Some high-grade flowers in greenhouse have very high demand on temperature and relative humidity. As the greenhouse is a system with great inertia, appropriate control methods must be adopted ahead in order to guarantee the temperature and relative humidity in reasonable scopes. The accurate prediction of temperature and relative humidity is the important prerequisite for adopting appropriate control methods. A combined prediction model was proposed based on the discrete grey model DGM(1,1) and linear model. In order to verify the validity of the DGM(1,1) model, the temperature and relative humidity were actually measured in a greenhouse. The results show that the largest relative error predicted by the combined model is 2.94% for temperature and 6.11% for relative humidity, which is superior to those of DGM(1,1) model and linear model only used respectively. At last, the factors affecting the prediction accuracy were analyzed. The change of control devices' working mode and the length of sampling period are two main influence factors.