Abstract
In this work, a new design of backpropagated intelligent computing paradigm is introduced to study the dynamics of the 3D nanofluid rheology along with a generalized heat flux model by exploiting the knack of neural network modeling supported with Levenberg-Marquardt backpropagation (NNLMB). The governing PDEs of the nonlinear fluid flow problem are converted to a system of ODEs by introducing appropriate similarity transformation. A dataset to implement the designed NNLMB is produced for variants of generalized heat flux model by variation of the thermophoresis parameter, Brownian motion parameter, Prandtl number, concentration relaxation parameter, Schmidt number, thermal relaxation parameter, temperature ratio parameter, power indices, Deborah number, and stretching ration parameter by using competency of the Adams numerical method (ANM). The dataset of the system model is arbitrarily segmented into training, testing, and validation samples to execute the NNLMB for finding the approximate results and comparison of obtained solutions is carried out with reference to access the accuracy of the designed NNLMB scheme. The worth of the design computing paradigm NNLMB is invariably endorsed with good agreements of the outcomes from reference ANM solutions with an accuracy of the order 10-05-10-08 for each scenario of generalized heat flux mode.