Levenberg–Marquardt neural network prediction of MHD hybrid nanofluid flow between parallel plates: Aerospace applications

Thermal performance is an important and common issue for today’s aeronautic technology and electrically powered devices. The following problems are commonly encountered due to increased heat generation or decreased surface area for cooling (limitation). One possible way to overcome this exciting challenge is to create a suitable cooling system architecture or to enhance the thermal transfer process. In this regard, hybrid nanofluids can help resolve these problems. In this work, the machine learning-based numerical analysis of MHD hybrid nanofluid flow through a rocket engine nozzle considering the effects of radiation, viscous dissipation and porous media is proposed. The hybrid nanofluid is produced by mixing basic fluid (kerosene oil) with the nanoparticles Ag-SiO2-TiO2 (Ag- silica- titanium dioxide), which have been chosen for their outstanding thermo-physical properties. Similarity transformations are used to define and solve the non-linear equations of momentum and energy and to guarantee stability and accuracy over a range of parametric regimes in the MATLAB bvp4c scheme. Considering important factors like Reynolds number (Re) and their effect on temperature distribution, skin friction, Nusselt number and velocity, the radiation, magnetic interaction, Eckert number and porosity are studied. When the thermal radiation and Eckert number values are rising, the thickness of the thermal boundary layer. The Levenberg–Marquardt technique optimises an Artificial Neural Network (ANN) to improve prediction capabilities. The ANN shows excellent agreement with numerical data, with correlation coefficients close to unity. The framework that has been suggested emphasises the significance of using hybrid nanofluids and data-driven methodologies to enhance rocket cooling efficiency and guarantee thermal stability in aerospace propulsion systems.