Application of Artificial Intelligence in Computational Fluid Dynamics

This review discusses the recent application of artificial intelligence (AI) algorithms in five aspects of computational fluid dynamics: aerodynamic models, turbulence models, some specific flows, and mass and heat transfer. Currently, there are three main coupling models. The first is the data-driven model to obtain the input-output relationship without involving any physical mechanisms. The second is the physical model to optimize the existing models by AI algorithms. The third is the hybrid model involving both data and physical mechanisms. Among various AI algorithms, artificial neural network is usually applied to build data-driven models and has been successfully employed in the mentioned five fields. Other AI algorithms such as recursive neural network, support vector machine, and naive Bayes are mainly used for the physical models. Finally, the development tendency of coupling models and how to choose an appropriate model are given in the conclusions and prospects.

Automated summary

This review discusses the recent application of artificial intelligence algorithms in computational fluid dynamics, focusing on data-driven models, physical models, and hybrid models. Among various AI algorithms, artificial neural networks are commonly used for building data-driven models. The conclusion also mentions the development tendency of coupling models and how to choose an appropriate model.