If you are interested in applying deep-learning tools to fluid mechanics and turbulence, perhaps you can find good ideas in my 2020 seminar delivered for the Australasian Fluid Mechanics Series. Here we discuss how to use recurrent neural networks (RNNs) for temporal predictions in turbulence, and convolutional neural networks (CNNs) for non-intrusive sensing in a turbulent channel.
You can find the recording below: