Characterization of a Swirled Flow Using Artificial Neural Networks

Abstract

Throughout history, humans have strived to enhance transportation and energy efficiency while mitigating environmental damage. The discovery of vortex flow in combustion technology has been pivotal, leading to ongoing research into its properties, especially in terms of the shape of the rotational areas it forms. This study delves into the use of artificial intelligence to predict vortex flow properties. Using experimental data, including descriptive and positional information, as inputs, and horizontal, vertical and kinetic energy as outputs across different locations within the combustion chamber, the model effectively captures the spatial features of the swirl flow field. It accurately predicts the velocity density distribution and vortex center position, which is in good agreement with experimental results. Furthermore, the generated prediction model shows promising accuracy over previous data sets, successfully reconstructing the vortex flow field and making inductive predictions on new data with a certain degree of generalizability. Ultimately, this study underscores the potential for many engineering applications to benefit from the prediction model developed here.