Numerical treatment of squeezing unsteady nanofluid flow using optimized stochastic algorithm
Ahcene Nouar, Amar Dib, Mohamed Kezzar, Mohamed Rafik Sari, Mohamed R. Eid
Abstract
Abstract In this paper, very efficient, intelligent techniques have been used to solve the fourth-order nonlinear ordinary differential equations arising from squeezing unsteady nanofluid flow. The activation functions used to develop the three models are log-sigmoid, radial basis, and tan-sigmoid. The neural network of each scheme is optimized with the interior point method (IPM) to find the weights of the networks. The confrontation of the obtained results with the numerical solutions shows good accuracy of the three schemes. The obtained solutions by utilizing the neural network technique of our variables field (velocity and temperature) are continuous contrary to the discrete form obtained by the numerical scheme.