Artificial neural network analysis for entropy optimized flow of Jeffrey nanofluid invoking Cattaneo-Christov theory

At present the study of artificial neural networks (ANNs) provide phenomenal applications such as data visualization, pattern recognition, predictive modelling, non-linear data processing, optimization control etc. Moreover Levenberg-Marquardt (LM) algorithm is utilized based on advanced computational approach of artificial neural networks (ANNs) in order to obtain complex non-linear solutions of heat and mass transfer for flow problems. Here chemically reactive flow of Jeffrey nanofluid followed by Cattaneo-Christov fluxes is analyzed. Flow is produced by curved stretched surface. Computational results and stability analysis of Jeffrey nanofluid are provided through neural network (ANN) approach including performance of validation sets to boost up system's accuracy, training state (TS) to optimize targeted datasets, regression analysis (R) to predict perfect output, error histogram (EH) to indicate the difference between targeted and predicted values and the best-fit to evaluate effectiveness of the system. New concept of entropy generation with Buongiorno's model and heat generation is discussed. Heat generation with Cattaneo-Christov flux model is considered in energy expression. Brownian movement and thermophoresis are also taken into account to discuss the properties of nanofluid. First-order chemical reaction is carried out for concentration equation. Mixed convective flow is observed. Transformations are utilized to alter non-linear system of partial differential expressions into non-dimensional ordinary expressions. Results of obtained nonlinear systems are numerically presented through utilizing bvp4c technique via MATLAB and then for obtaining more advanced and comparative solutions the predictive capabilities of advanced neural networks (ANNs) algorithm is employed to train the optimized datasets. Graphical interpretations of temperature, entropy generation, flow field and concentration through influential variables are analyzed. Clearly one can find that higher thermal relaxation time variable leads to entropy generation enhancement. Velocity field rises for larger Deborah number. Reduction in concentration through larger solutal relaxation time variable is noticed. A detailed inspection of comparison between ANN approach and numerical technique (bvp4c) is observed through perfect match of their outcomes.