Prediction of the thermophysical properties of Ag-reduced graphene oxide-water/ethylene-glycol hybrid nanofluids using different machine learning methods

Because of their enhanced thermophysical characteristics, namely greater thermal conductivity, viscosity control, and long-term stability than traditional nanofluids, hybrid nanofluids drew interest. Such properties make them suitable candidates for many industrial applications such as solar systems and thermal management. However, knowing the thermophysical properties of these materials accurately is difficult because of the complexities of nanoparticles and the interaction with the base fluid. This paper utilizes machine learning methods to predict the thermophysical properties of water/ethylene glycol mixture-based hybrid nanofluids containing reduced silver-graphene oxide. ology: This study aimed to predict Viscosity (DV), Thermal Conductivity (TC) and Density (D) by three machine learning algorithms including multiple linear regression (MLR), Multiple Polynomial Regression (MPR) and Gaussian Process Regression (GPR). A 5 × 28 dataset was used for training and testing the network, with 80 % of the data used for training the network and 20 % for testing the network. Evaluating the performance of algorithms is based on the evaluation indices of Correlation coefficient (R), Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Standard Deviation (STD). In addition, optimization is done by the Non-dominated Sorting Genetic Algorithm-II (NSGA-Ⅱ) algorithm and the impact results of different mutation and combination rates are examined. The MPR algorithm yielded the lowest MoD values (0.07 % and −0.06 %) and the highest prediction accuracy among the models tested (R = 0.9999, RMSD = 2.726 × 10 −4 , STD = 0.0219). Furthermore, NSGA-II optimization results revealed that the temperature and concentration of nanoparticles could effectively increase the thermal conductivity, while too high concentration could also increase viscosity. Finally, through the TOPSIS method, the best point was chosen giving a blend of ideal thermophysical properties. This signifies that machine learning methods can be successfully employed for the prediction and optimization of hybrid nanofluid characteristics. • Using a machine learning system to predict the results of laboratory experiments is the stated goal of this effort. • The 5∗28 dataset is fed into the algorithm, and the last three columns display the objective function or output: • The data is used by the machine learning algorithm to generate predictions. • Predicted points are both quite precise and well-matched with experimental placements. • With the lowest MoD values of 0.07 % and −0.06 %, the MPR model seems to offer the greatest mix of accuracy.