Heat transfer and MLP neural network models to predict inside environment variables and energy lost in a semi-solar greenhouse

Abstract

The greenhouse environment is an uncertain nonlinear system which classical modeling methods have some problems to solve. There are many control methods, such as adaptive, feedback and intelligent control and they require a precise model. Therefore, many modeling methods have been proposed for this purpose, including physical, transfer function and black-box modeling. The main goal of this paper is to compare some mathematical models (include dynamic and multiple linear regression (MLR)) with innovative method (Artificial Neural Network) and select the best one to predict inside air and roof temperature (Ta and Tri) and energy lost in a semi-solar greenhouse in Iran. For this purpose, a semi-solar greenhouse was designed and constructed at the North-West of Iran in Azerbaijan Province (geographical location of 38°10′N and 46°18′E with elevation of 1364m above the sea level). The environment factors influencing the Ta and Tri include outside air temperature (To), wind speed (vo), solar radiation on the roof (Io), inside soil temperature (Ts) and inside air humidity (RHa), which were all collected as data samples. Then through the relationship between the factors, 4 main factors were extracted, and the relationship between the main factors and the original data was discussed by MLP and MLR models. Results showed that the Durbin–Watson statistic for MLR method to estimate Ta and Tri was 0.04 and 0.06 respectively, so this method cannot predict the output parameters correctly. Comparing MLP and dynamic models showed that the performance of MLP model was better according to small values of RMSE and MAPE and large value of EF indices. Statistical comparisons of the predicted data by neural network models and the actual data of the inside air and roof temperature showed that there is no significant difference between them. Also, the minimum value of the TSSE was 16.68 and 30.87 (°C2) for Ta and Tri in ANN implementation. The performance of best model (MLP) to estimate the energy lost and exchange in a semi-solar greenhouse showed that MLP method is applicable to estimate the real data in greenhouse and then predict the energy lost and exchange. This method can be used online in greenhouses to decrease some cost related to application of sensors and some record instruments.