Application of support vector machine for prediction of electrical and thermal performance in PV/T system

Abstract

In photovoltaic–thermal (PV/T) system analysis, solar collectors with numerous design concepts have been used to purvey the thermal and electrical energy effectively. In this study, two types of solar thermal collectors in PV/T system are proposed and fabricated called design A and design B respectively. In order to investigate the effects of collector type on the system performance a thin flat metallic sheet (TFMS) and fins were introduced as an effective heat absorber and heat sink in the collectors. Extensive experiments were carried out for different conditions under indoor solar simulator. Then PV/T thermal and electrical efficiency were calculated by using data obtained from experiments. Here, support vector machine (SVM) model is designed to estimate the thermal and electrical output which predicts the values for some input variables. For this purpose, three SVM models namely SVM coupled with the discrete wavelet transform (SVM-Wavelet), the firefly algorithm (SVM-FFA) and with using the radial basis function (SVM-RBF) were analyzed. The estimation and prediction results of these models were compared with each other using statistical indicators i.e. root means square error, coefficient of determination and Pearson coefficient. The experimental results show that a significant improvement in predictive accuracy and capability of generalization can be achieved by the SVM-Wavelet approach. Moreover, the results indicate that proposed SVM-Wavelet model can adequately predict the electrical and thermal efficiencies of PV/T system. In the final analysis, a proper sensitivity analysis is performed to identify the influence of considered input elements on performance prediction of PV/T system.