Intelligent computing applications to study the tri-hybrid nanofluid past over the stretched surface

• The study employs Recurrent Neural Networks optimized by the Levenberg-Marquardt method (RNN-LMB) to examine the dynamics of the hybrid nanofluid model THNF-MHDSP. • The similarity transformations are employed to transform the complex PDEs that describe THNF-MHDSP into simpler ODEs. • Variables such as velocity slip parameter ( β ) , magnetic parameter M, solid volume fraction ( φ 3 ), suction parameter (S), viscosity parameter (θr), and Eckert number (Ec) are analyzed by creating a dataset for RNN-LMB using the ND solve method across different versions of THNF-MHDSP. • The approximate solutions were obtained with 6–8 decimal places of accuracy and demonstratd for the THNF-MHDSP model by the training, testing and validation processes of the developed RNN-LMB scheme. • The accuracy of THNF-MHDSP is confirmed through analysis of Mean Squared Error curves, error histograms, and various statistical tests, consistently proving the solver’s reliability. In this study, a recurrent neural network with the Levenberg-Marquardt backpropagation algorithm is used to examine the effects of various parameters on cooling mechanisms involving tri-hybrid nanofluid consisting of Al 2 O 3 , Cu , TiO 2 and magnetohydrodynamic stagnation point flow. This research focus on nanofluid in the existence of an asymmetrically stretched disc in water, which plays an imprortant role for understanding heat transfer in addition to the dynamics of momentum and thermal boundary surface. Transform the complex partial differential system through suitable similarity transformations into the ordinary differential equations for numerical computing. These equations have been solved numerically by using the Adams method in Mathematica to generate datasets for various parameters. The current study demonstrates that the velocity profile rises with higher magnetic, velocity slip, and mass suction parameters. The fluid thermal level tends to decrease with an increase in mass suction parameters, whereas an increase in the magnetic parameter and Eckert number raises the fluid’s thermal level. Furthermore, it is noticed that the flow resistance and internal friction of the fluid are proportional to the viscosity parameter. The proposed scheme has shown its effectiveness in a wide range of scenarios and cases. This was achieved by comparing the mean square error metric for the squeezing flow tri-hybrid nanofluid problem with the obtained results. In addition, various statistical measures like state transitions, fitness assessments, error histograms, and regression analyses, provide further evidence of the solver’s robustness and reliability. The excellent agreement between the reference datasets and refined numerical solutions through the scheme showcases the method’s effectiveness, achieving remarkable accuracy levels ranging from 10 -6 to 10 -12 .