Artificial neural network-based study of entropy optimization in Johnson-Segalman nanofluids through a peristaltic channel

作者：Muhammad Ishaq, M. Bilal Ashraf, Moieza Ashraf, Sultan Alshehery, Abdullah A. Faqihi, Haitham Hadidi · 发表于：Physics of Fluids · 年份：2025 · DOI：10.1063/5.0255518 · 被引用次数：28 · 研究领域：Nanofluid Flow and Heat Transfer、Heat Transfer and Optimization、Rheology and Fluid Dynamics Studies

This study includes an artificial neural network (ANN) analysis of irreversibility in Johnson–Segalman nanofluid flow through a peristaltic channel under the influence of motile microorganisms, viscous dissipation, and slip effects. The nonlinear partial differential equations are transformed into ordinary differential equations by applying the lubrication approximation and Debye–Hückel transformations with the help of suitable dimensionless variables. The resultant dimensionless ordinary differential equations are solved analytically using the homotopy perturbation method (HPM) by linearizing and assuming a series solution. The linear subproblems from HPM are solved successively to find the symbolic series solution in MATLAB by utilizing the dsolve command. The symbolic solutions for velocity, temperature, concentration, and bioconvection are plotted against different physical parameters to visualize their behavior and profiles. Moreover, data for velocity, thermal, concentration, and bioconvection profiles are extracted to train the ANN model. The ANN model is trained in Python using TensorFlow version 2.17.0., and it consists of one input layer, two hidden layers (each with 64 neurons), and one output layer. The ReLU activation function is used in the hidden layers, and the Adam optimizer is employed in our model. Performance metrics such as mean square error (MSE), regression (R2), error histogram, gradient, and relative error, and absolute error are computed to monitor t...