Forecasting the Outlet Fluid Temperature from a Flat Plate Collector Using Artificial Neural Networks (ANNs) and Support Vector Regression (SVR)

In the present study, the outlet water temperature from flat plate solar collector using artificial neural networks (ANNs) and support vector regression (SVR) was modeled and compared with experimental results. Based on the results, with increasing input parameters of models, the accuracy of the model was increased. According to the results the values ​​of R2, RMSE and MAPE in the SVR method for the first model were 0.97, 3.25 and 2.77, respectively. While these values for the second model was 0.99, 0.10 and 0.55, respectively. On the other hand, ​​for the ANN method and for the first model these values were 0.99 and 0.02 and 0.28, respectively. And for the second model were 0.99 and 0.01 and 0.19, respectively. The results showed that the accuracy of artificial neural network model for peridicting the water outlet temperature was better than that of the support vector regression model.