Prediction of Temperature in a Greenhouse Covered with Polyethylene Plastic Using Artificial Neural Networks, Case Study: Jiroft Region

Internal temperatures of greenhouse and its control is one of the important parameters in greenhouses and plays a key role in the economics of production. Although the greenhouse is a closed environment, it is not completely isolated from the outside. Therefore, the conditions inside the greenhouse are constantly changing under the influence of outside climate change. The purpose of this study was to estimate the internal air temperature of polyethylene greenhouse with respect to the external parameters of the greenhouse including air temperature (Tout), air relative humidity (Hout), solar radiation (S) and wind speed (V). For this purpose, different method of artificial neural networks including Multilayer Perceptron (MLP), Radial Basis Function (RBF) and Adaptive Nero Fuzzy Inference System (ANFIS) were used. Comparison between different neural network models showed that RBF method had better prediction performance than MLP and ANFIS with higher coefficient of determination (R2=0.93) and lower error (RMSE=2.25). The results of the RBF model estimation for the prediction future temperature indicated an acceptable error in the prediction by the model for the next two hours and thus, the farmers had enough time to provide the necessary measures to prevent the greenhouse temperature rise in the future and save in energy consumption.