Artificial Neural Network Experimental Data Prediction of a Long-wave Excited Plane Jet Part I: Near Field Comparison

Abstract

Present study used artificial neural network method to investigate the flow structure development of a low speed plane jet based on averaged and instantaneous velocity got by hot-wire measurement. Especially the mode transformation from sub-harmonic mode at upstream to low-frequency flapping mode at downstream. All experimental results at this stage all show a small difference and error from the actual value. The overall MAPE scale also reaches the standard of 10^(-3) and indicates that the average error is quite low. Base on the spectrum analysis, it can be known that the instantaneous velocity obtained by simulation can fully express the mode of the actual flow field. The built numerical simulation model can be applied to predict the velocity signal and know the development of the near field flow structure and mode development.