Artificial neural network and application at Ranque-Hilsch vortex tube

Artificial Neural Networks are one of the fields which have shown great interest by artificial intelligence researchers. Artificial Neural Networks; a very successful tool for solving problems, thanks to features such as self-learning or learning from examples and ability to make generalizations are spreading very rapidly and intensively used in researches, applications due to technological developments in recent years. Artificial Neural Networks are especially preferred for non-linear, difficult to modelling mathematical problems due to conveniences provided for to solve the problems.Scope of this thesis is; to practice counter-flow Ranque-Hilsch vortex tube by using experimental data. Primarily, Artificial Neural Networks were theoretically explained and it?s structure, basic learning methods were mentioned. For the generated network, multilayer perceptron model was used and the performance of the vortex tube was modelled with neural networks.As a result of this thesis, to observe how transfer function, data selection method and number of data effects to Artificial Neural Network?s performance.