A Data Augmentation-Based Technique for Deep Learning Applied to CFD Simulations

The computational cost and memory demand required by fluid dynamics (CFD) codes simulations can become very high. Therefore, the application of convolutional neural networks (CNN) in this field has been studied owing to its capacity learn patterns from sets input data, which considerably approximate results CFD with relative low errors. DeepCFD code taken as a basis some slight variations parameters CNN, while net is able solve Navier–Stokes equations for steady turbulent flows variable velocities domain. In order acquire extensive data augmentation technique, considers similarity principle dynamics, implemented. As consequence, pressure fields quite accurately, speeding up time-consuming simulations.