Intelligent Prediction on Mixed Convection Flow via Neural Networks

Dedeoglu A. N., Pekmen B.

7th International Conference on Intelligent and Fuzzy Systems, INFUS 2025, İstanbul, Turkey, 29 - 31 July 2025, vol.1529 LNNS, pp.271-278, (Full Text) identifier

  • Publication Type: Conference Paper / Full Text
  • Volume: 1529 LNNS
  • Doi Number: 10.1007/978-3-031-97992-7_31
  • City: İstanbul
  • Country: Turkey
  • Page Numbers: pp.271-278
  • Keywords: lid- driven cavity, machine learning, mixed convection flow, neural network, RBF method
  • TED University Affiliated: Yes

Abstract

In this study, mixed convection flow in a lid driven cavity is investigated both by numerical and machine learning (ML) approaches. In the numerical part, the problem is solved by radial basis function (RBF) method in space derivatives and backward differentiation formula of order two (BDF2) in time derivatives. In the ML part, neural network (NN) modeling is utilized using the data obtained from the numerical results to model some important problem indicators as average Nusselt number, temperature variance and absolute maximum stream function value. The importance of Reynolds number on average Nusselt number and temperature variance, and of Grashof number on absolute maximum stream function are noted. The predicted results are also found in good harmony with the numerical results in terms of mean squared error metric results. The NN modeling enables one to interpret the thermal and fluid behavior of the considered system immediately at the desired problem parameters instead of running the numerical calculations many times.