Comparative Performance Analysis of Support Vector Machine Models for Predicting THPP of Solar Parabolic Trough Receiver with Twisted Tape Inserts

Abstract

This study aims to compare the performance of different Support Vector Machine (SVM) models, namely linear SVM, quadratic SVM, cubic SVM, fine Gaussian SVM, medium Gaussian SVM, and coarse Gaussian SVM, for predicting the thermal hydraulic performance parameter (THPP) of a solar parabolic trough receiver with twisted tape inserts. The dataset used in this study was obtained through numerical simulations and comprises different configurations of twisted tape inserts, flow rates, and solar irradiation levels. The experimental design included three twist ratios (3, 4, and 5), five perforation ratios (0.05, 0.1, 0.15, 0.20, and 0.25), and three wing depth ratios (0.1, 0.2, and 0.3), resulting in a total of 210 data sets. The SVM models were trained using the dataset and evaluated based on their accuracy and computational efficiency. The input layer of the study consisted of both flow and geometrical parameters. The flow parameters included the mass flow rate, outside temperature, average temperature of the twisted tape, and average temperature of the air on the film. The geometrical parameters included the twist ratio, wing depth ratio, and perforation ratio of the twisted tape insert. The model's performance is evaluated using Root Mean Square Error (RMSE) and the coefficient of determination (R2). For the cubic SVM, it has been discovered that the values of RMSE and R2 are 0.054443 and 0.98 respectively. The results showed that the cubic SVM model had the best performance in terms of accuracy, followed by the quadratic SVM, medium Gaussian SVM and other models. These findings provide valuable insights into the selection of SVM models for predicting the THPP of solar parabolic trough receivers with twisted tape inserts.