jump to content

jump to main navigation

TU Berlin

Document content

There is no English translation for this web page.

Compact Lecture WS 19/20 Mathematical Methods of Turbulence Control I:

Tools of Machine Learning for the Tasks of Fluid Mechanics

Prof. Bernd R. Noack (www.berndnoack.com)

March 2nd-6th (Monday- Friday) 2020, 9:15-12:00 h
Room: BIB014

Machine learning (ML) / artificial intelligence (AI) approaches are penetrating almost every aspect of our everyday life.

Machine learning also rapidly revolutionizes research in fluid mechanics. First principle approaches of theoretical and computational fluid mechanics are increasingly augmented - if not replaced - by data-driven methods - from interpolations using abundant data to „clever guesses“ in case of the lack thereof. One reason is the enormous challenge posed by the comlexity of high Reynolds number flows, by a large variety of operating conditions and by uncertainty, to name a few.

This course aims to give a systematic introduction to data-driven fluid mechanics. Most tasks in fluid mechanics can be formulated as regression problems, i.e. finding a mapping which minimizes a cost. For instance, given the current state, predict the future state as good as possible or given a flow plant, find a control law optimizing a performance. Many methods of machine learning solve regression problems. This course will provide a swiss army knife of machine learning tools for fluid mechanics tasks. First, six classical regression problems of fluid mechanics are introduced. Then, four types of regression problems are distinguished. Finally few dozen regression solvers (first principles and ML) are introduced with their relative strengths and weaknesses.

The students are encouraged to bring their own research problems for a discussion in class and for a potential one-to-one discussion in the afternoon. While the first course of this module details proven strategies of turbulence control, this second course is devoted to the tools or “building blocks” used for control, modeling, analysis and optimization. Literature: Pedro Domingos 2018 The Master Agorithm: How The Quest For The Ultimate Learning Machine Will Remake Our World, Basic Books, NY, USA S.L. Brunton & B.R. Noack 2015 Appl. Mech. Rev. (provides a 48 page overview)


Contact & Registration: Christian Navid Nayeri

Auxiliary Functions / Extras

Help Functions