Python-based deep learning for optimizing thermal performance of Prandtl-Eyring hybrid nanofluids in solar systems

Improved heat transfer efficiency is greatly needed for the advancement of sustainable solar energy systems. Hybrid nanofluids could present strong options for maximizing thermal performance due to improved thermophysical properties. This study presents a Python-based hybrid framework integrating the machine learning deep learning neural network (DLNN) model and a boundary value problem solver to examine and determine the thermal behavior for Prandtl-Eyring hybrid nanofluids in conditions of solar heating. Using engine oil as a base fluid and Cu–MoS 2 nanoparticles at a 2% volume fraction, the framework utilized the solver that will be used along with solve_bvp to find a solution of the transformed ODEs, and the model will train a DLNN (two hidden layers, with ReLU activation and optimized with Adam) to have the ability to conduct predictive analysis. The hybrid nanofluid, by itself, outperformed the mono nanofluid's capacity for heat transfer efficiency, with a small mean-squared error across the simulations, in addition to improved thermal profiles. The framework and methodology here presented are important and innovative for optimizing solar water pump systems for off-grid agricultural irrigation applications and clean energy distribution.