Predicting Turbulent Systems from Limited Measurements: Classical Methods to Machine Learning
报告人：Vikrant Gupta 研究副教授（南方科技大学）
Recent advancements in measurement techniques and computational power have enabled the use of data assimilation and neural networks for turbulence estimation and forecasting. However, there is limited understanding of the measurement data needed to accurately reconstruct and predict flow fields. In this seminar, I will analyse three main predictive methods: (i) linearised (low-rank approximation) models, (ii) data assimilation and (iii) model-free neural networks. On the one hand, linearised models require little measurement data but necessitate deep understanding of system dynamics. On the other hand, model-free networks require high-resolution space-time measurement data but require little to no knowledge of system dynamics. I will demonstrate that the methods within each class have similar measurement data, which are closely related to measures from chaos theory. These findings can guide the systematic collection of data and selection of predictive methods for turbulence forecasting in practical systems.
Dr. Vikrant Gupta is currently a Research Associate Professor at the Southern University of Science & Technology where he applies data-driven and dynamical systems tools to study complex flows. The areas of application include wall turbulence, wind and tidal energy, and low-emissions gas turbines. He has obtained his PhD from the University of Cambridge for which he was awarded a Dorothy Hodgkin Postgraduate Award and his bachelor’s and master’s degrees from the Indian Institute of Technology Madras. He has published 24 SCI papers in high-impact-factor journals, including 10 in Journal of Fluid Mechanics. He has also been awarded two NSFC grants for his research on wall turbulence and wind energy.
Center for Applied Physics and Technology,Peking University
Room402,1 Building,College of Engineering ,PKU