The role of neural networks in fluid mechanics and heat transfer

The recent applications of Artificial Neural Networks (ANNs) to fluid mechanics and heat transfer are presented. ANNs have proved beneficial by their capability in modelling complex nonlinear problems as well as providing a fast, automatic method in some applications. In heat transfer the backpropagation model has been predominant and it has also been widely used in fluid mechanics. However, flow visualization has witnessed a substantial application of unsupervised learning algorithms. Finally, a novel technique that uses two ART2 (Adaptive Resonance Theory) networks to determine fluid flow velocities has been developed by the authors resulting in accuracies of up to 96.4%