Advanced thermal transmission using tetrahybrid nanofluids: numerical simulation and soft-computing-based strategic perspectives

One of the most important and sustainable sources of thermal energy is solar energy, which is produced by the Sun. Numerous technological applications, such as photovoltaic systems, renewable power generation, solar street lighting, and solar-powered water pumps, have emerged from its immense potential. The use of solar energy has expanded into more sophisticated industrial fields in the modern era, including technological advancements in solar-powered vehicles. To examine the heat transmission capabilities of photovoltaic hybrid automobiles, this study presents a novel computational framework that combines Artificial Intelligence (AI) methods with a numerical modelling strategy. The Casson–Sutterby Tetrahybrid Nanofluid Model (CSTNFM), which considers the combined impacts of tetrahybrid nanoparticles, internal heat generation, thermal radiation, and variable thermal conductivity, serves as the basis for the analysis. The nanofluid is a four-component mixture of SiO2, Al2O3, TiO2, and Cu nanoparticles, with ethylene glycol serving as the base fluid. To evaluate parameters such as skin friction coefficient, heat transfer rate, velocity, and temperature profiles, the bvp4c solver in MATLAB® is used to solve the governing nonlinear boundary value problems. Using a neural fitting framework, the tabulated data is processed further. A feed-forward Neural Network Backpropagation Levenberg–Marquardt Technique (NNB-LMT), is adopted. Training, testing, and validation datasets are divided into 70%, 15%, and 15%, respectively. Through successive training, validation, and testing within the NNB-LMT framework, the proposed hybrid numerical–AI methodology generates performance plots, state functions, error histograms, regression analyses, and fitting curves analysis. The reliability of the suggested approach is confirmed by the generated neural model's ability to approximate the CSTNFM solutions in various circumstances, yielding results that closely match the reference data. Regression and histogram analyses confirm that the AI-driven neural method has outstanding convergence behaviour, with mean squared error ranging from 10−09 to 10−06. These results validate the AI-integrated thermal transport model's high accuracy and resilience [Q1].