On the specific heat capacity estimation of metal oxide-based nanofluid for energy perspective – A comprehensive assessment of data analysis techniques

Abstract The main aim of the present study is to investigate the capabilities of four robust machine learning method - the Kernel Extreme Learning Machine (KELM), Adaptive Regression Spline (MARS), M5 Model Tree (M5Tree), and Gene Expression Programming (GEP) model in predicting specific heat capacity (SHC) of metal oxide-based nanofluids implemented in solar energy application. Sets of 1180 data of different metal oxide-based nanofluids containing Al2O3, ZnO, TiO2, SiO2, MgO, and CuO dispersed in various base fluids were collected from reliable literature to provide the predictive model of SHC of nanofluids. The volume fraction, temperature, SHC of the base fluid, and mean diameter of nanoparticles were used as an input variable to predict nanofluids' SHC as the output variable. The artificial intelligence (AI) models were validated using several statistical performance criteria, graphical devices, and conventional models. The results obtained from all datasets demonstrated that the KELM model significantly outperformed the MARS, M5Tree, and GEP model in predicting the SHC of nanofluid. Moreover, the sensitivity analysis showed that the mean diameter of the nanoparticle and SHC of the base fluid have the most considerable impact on estimating the SHC of metal oxide-based nanofluids.