Physics Driven Machine Learning Surrogates for Applied Aerodynamics
The resurgence of Machine Learning (ML) since 2012 has revealed visible impact in many fields, resulting in several useful products and processes shaping the Industry 4.0 agenda. This has caught the attention of computational scientists, who for the past half century or so have advanced computational continuum physics driven methods tools for the analysis and synthesis of engineered systems, facilitated primarily by fast and cheap high performance computing platforms, efficient numerical algorithms and innovative software systems. Since 2017, computational scientists have begun exploring ML in computational science and engineering and as recent as 2019/2020, in establishing an agenda for scientific machine learning which aims to build ML surrogates for predicting system performance and characteristics of engineered systems rapidly and realistically to increase productivity. This seminar will outline the prospects for scientific machine learning surrogates as alternative computational prediction tools for various problems in applied aerodynamics and their role in constructing digital twins for aerospace systems.
Who Should Attend
Dr. Murali Damodaran
Dr. Murali Damodaran is currently a Senior Research Scientist with the Centre for Aerodynamics and Propulsion at Temasek Laboratories, NUS, Singapore (since 2018). Prior to this, he was on the faculty of Mechanical and Aerospace engineering at NTU, Singapore (1989-2010), IIT Gandhinagar, India (2010-2015) and NUS (2016-2018). His research interests are in the broad areas of Aerospace and Computational Engineering and of late has been exploring possibilities of machine learning in scientific computing for engineering applications in the wake of the AI resurgence. He has a BTech in Aeronautical Engineering from IIT Kanpur, India and a MS and PhD in Aerospace Engineering from Cornell University, USA.
His profile can be found at https://temasek-labs.nus.edu.sg/program/program_lowspeedaerodynamics_tslmura.html.