Ricardo Vinuesa on Explainability framework & Deep RL for turbulent flow control (FSML Seminar 08) скачать в хорошем качестве

Ricardo Vinuesa on Explainability framework & Deep RL for turbulent flow control (FSML Seminar 08)

Many thanks to Dr. Ricardo Vinuesa for this great talk. This was given as part of the Frontiers in Scientific Machine Learning Seminar Series at the University of Michigan. Check out some of Dr. Vinuesa's other work and explainers on his channel! -    / @rvinuesa   Abstract: In this work we first use explainable deep learning based on Shapley explanations to identify the most important regions for predicting the future states of a turbulent channel flow. The explainability framework (based on gradient SHAP) is applied to each grid point in the domain, and through percolation analysis we identify coherent flow regions of high importance. These regions have around 70% overlap with the intense Reynolds-stress (Q) events in two-dimensional vertical planes. Interestingly, these importance-based structures have high overlap with classical turbulence structures (Q events, streaks and vortex clusters) in different wall-normal locations, suggesting that this new framework provides a more comprehensive way to study turbulence. We also discuss the application of deep reinforcement learning (DRL) to discover active-flow-control strategies for turbulent flows, including turbulent channels, three-dimensional cylinders and turbulent separation bubbles. In all the cases, the discovered DRL-based strategies significantly outperform classical flow-control approaches. We conclude that DRL has tremendous potential for drag reduction in a wide range of complex turbulent-flow configurations. Bio: Dr. Ricardo Vinuesa is joining the Department of Aerospace Engineering at the University of Michigan in the Fall of 2025. He is currently an Associate Professor at the Department of Engineering Mechanics, KTH Royal Institute of Technology in Stockholm. He studied Mechanical Engineering at the Polytechnic University of Valencia (Spain), and he received his PhD in Mechanical and Aerospace Engineering from the Illinois Institute of Technology in Chicago. His research combines numerical simulations and data-driven methods to understand, control and predict complex wall-bounded turbulent flows, such as the boundary layers developing around wings and urban environments. Dr. Vinuesa has received, among others, an ERC Consolidator Grant, the TSFP Kasagi Award, the MST Emerging Leaders Award, the Goran Gustafsson Award for Young Researchers, the IIT Outstanding Young Alumnus Award, the SARES Young Researcher Award and he leads several large Horizon Europe projects. He is also a member of the Young Academy of Science of Spain. Sections: 00:00 Introduction 02:38 Motivation, AI4Science 06:28 Part 1: Defining important regions in the flow 08:42 Part 1: XAI Framework for Important Coherent Flow Structures 14:40 Part 1: Interpretation of SHAP Values 18:29 Part 1: Identifying new coherent structures with Gradient SHAP 26:05 Part 1: Combining autoencoders, causality and SHAP 30:30 Part 2: Deep Reinforcement Learning (DRL) for Flow Control 33:19 Part 2: Multi-agent Reinforcement Learning (MARL) 37:18 Part 2: Quadrant Analysis 39:52 Part 2: Targeting most important coherent structures 42:24 Part 2: Application of DRL to other turbulent flows 45:17 Summary and Conclusions 46:28 Q&A Session