FOX-Inspired Optimizer with Support Vector Regression Modeling of Water-Based CaCO3-CuO-SiO2 Trihybrid Nanofluids Thermal Properties: Comparative Study

Recent advances in artificial intelligence have spurred significant interest in accurately predicting the thermophysical properties and rheological behavior of nanofluids. This study introduces four support vector regression (SVR) models optimized using the Dragonfly Algorithm (DA) and the novel FOX-inspired Optimization Algorithm (FOA). The models were evaluated with two cross-validation techniques, Leave-M-Out (LMO) and Holdout, to estimate the thermal properties of trihybrid nanofluids (THNFs). Trained and tested on a diverse dataset compiled from published experimental studies, these models exhibited exceptional predictive accuracy. Performance evaluation using metrics such as mean squared error (MSE) and Theil's U² revealed remarkably low error values, with all models achieving correlation coefficients (R) and determination coefficients (R²) exceeding 0.999. The results demonstrate the superior capability of these models to predict dynamic viscosity and thermal conductivity with high precision. This study’s findings hold substantial industrial significance, particularly in energy, thermal management, and manufacturing sectors.