Morlet-wavelet neural networks and sensitivity analysis in magnetized peristaltic flow subject to Soret–Dufour effects: An unsupervised approach

Zakir Hussain,Metib Alghamdi,Muhammad Naeem Aslam,Taseer Muhammad
DOI: https://doi.org/10.1016/j.icheatmasstransfer.2024.108259
IF: 6.782
2024-11-10
International Communications in Heat and Mass Transfer
Abstract:This study discloses peristaltic flow under the influence of an inclined magnetic field, Hall effect, Soret-Dufour effects, and convective conditions in an asymmetric inclined channel. The physical characteristics of nanomaterials are incorporated into the flow. A mathematical flow model has been developed for momentum, concentration, and energy, considering conservation principles. Such models are simplified through small Reynolds number and long-wavelength assumptions. Results for flow quantities are obtained using an unsupervised approach, the Morlet-wavelet neural network (MWNN) hybridized with Particle Swarm Optimization (PSO). An error-based fitness function is developed for the MWNN formulation, and the fitness function is optimized for decision variables in the MWNN using the PSO algorithm. Additionally, MWNN-PSO results are matched with published work and found to be in excellent agreement. Furthermore, Response Surface Methodology is used for sensitivity analysis. As a result, Hall effect, Dufour number, and the inclined magnetic variable show strong correlation and significant effects on heat transport. The findings of the study reflect that heat transfer enhances when Froude and Brinkman numbers increase, while an opposite trend is noted for the Dufour number. This study provides insights into identifying significant variables and optimizing heat transfer efficiency precisely.
thermodynamics,mechanics
What problem does this paper attempt to address?