Artificial intelligence-based analysis employing Levenberg Marquardt neural networks to study chemically reactive thermally radiative tangent hyperbolic nanofluid flow considering Darcy-Forchheimer theory
Hamid Qureshi, Usman Khaliq, Zahoor Shah, Hajar Abutuqayqah, Muhammad Waqas, S. Saleem, Waqar Azeem Khan
Abstract
The complexity of fluid dynamics, or, more specifically, the use of reacting, radiative emitting/absorbing chemically active nanofluids, poses a problem to conventional computational models. To overcome these difficulties, this research seeks to apply the Levenberg–Marquardt Neural Networks (LMA-NN) to analyze the Darcy-Forchheimer solution for flow of radiated tangent hyperbolic fluid using viscous dissipation and activation energy subject to parameter physical features. To convert the nonlinear PDEs into their equivalent ODEs, the similarity transformation coefficients are applied. For each variable, the resulting outcomes are explained. The use of AI is to approximate computations of various physical quantities using datasets generated by the ND-solver algorithm in MATHEMATICA environment. Further, the results are discussed for the variables: curvature parameter, Weissenberg number , Prandtl number, radiation parameter, Schmidt number and thermal diffusivity coefficients. Probabilistic instance distribution research involving error-histograms, continual curve modeling at each progressive step at epochs, review of the adaptive control parameters of artificial intelligence (AI) based feeding neural network computing, and testing of the coefficient through regression model metric are used to establish and validate the convergence, accuracy as well as the effectiveness of the recommended AI-based analysis on Levenberg Marquardt evaluator through MATLAB. The results in eight scenarios for the proposed Levenberg–Marquardt neural networks model were highly precise, with error values of 6.76E-12, 2.58E-10, 4.07E-10, 2.24E-10, 6.90E-11, 4.04E-10, 4.85E-10, and 1. The contrary epochs of convergence were 115, 436, 299, 564, 236, 332, 145, and 416, correspondingly. The above outcomes demonstrate the effectiveness of the idea in using the given problem in fluid dynamics analysis while also proving to be less erroneous than other existing powerful techniques.