Modelling and Controlling Turbulent Flows Through Deep Learning

Explore the application of deep neural networks (DNNs) in computational fluid dynamics (CFD) through this 52-minute lecture by Ricardo Vinuesa at the Alan Turing Institute. Delve into the fundamentals of deep learning in fluid mechanics and discover how convolutional neural networks (CNNs) can be used for non-intrusive sensing in turbulent flows. Learn about the superior performance of DNNs compared to traditional linear models in predicting flow in turbulent open channels based on wall measurements. Examine the potential of transfer learning between different friction Reynolds numbers. Investigate additional modelling techniques using autoencoders (AEs) and generative adversarial networks (GANs). Conclude by exploring the applications of deep-reinforcement-learning-based flow control in turbulent flow scenarios.