Vivek OOmmen: The effectiveness of PINNs in solving inverse heat transfer problems

Machine Learning Seminar presentation

Topic: The effectiveness of PINNs in solving inverse heat transfer problems

Speakers: Vivek Oommen & Prof. Balaji Srinivasan, University of Luxembourg, Department of Computational Science

Time: Wednesday 2020.12.02, 10:00 CET

How to join: Please contact Jakub Lengiewicz

Abstract:

In this presentation, I would like to demonstrate the effectiveness of PINNs in solving several inverse heat transfer problems. To justify its effectiveness, the time taken by PINNs to solve the entire problem is compared with the time taken by COMSOL Multiphysics Software (FEM approach) to solve the forward problem once. Steady and unsteady problems governed by both linear and nonlinear PDEs have been chosen as the test cases, to check if PINNs can handle a variety of problems often encountered in the field of heat transfer. I will conclude after mentioning a specific type of PINN architecture that works very well for PDEs that are highly nonlinear.

Additional material:

[1] “Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations” (M. Raissi, P. Perdikaris and G.E. Karniadakis), (Journal of Computational Physics)
[2] “Estimation of the non-linear diffusion coefficient with Markov Chain Monte Carlo method based on the integral information” (Zbigniew Bulinski and Helcio R.B. Orlande) (International Journal of Numerical Methods for Heat & Fluid Flow)
[3] “A Bayesian approach for the simultaneous estimation of surface heat transfer coefficient and thermal conductivity from steady state experiments on fins” (N. Gnanasekaran and C. Balaji), (International Journal of Heat and Mass Transfer)
[4] “Application of genetic algorithm for unknown parameter estimations in cylindrical fin” (Ranjan Das), (Applied Soft Computing)
[5] “Application of Adomian decomposition method and inverse solution for a fin with variable thermal conductivity and heat generation” (Rohit K. Singla and Ranjan Das), (International Journal of Heat and Mass Transfer)
[6] “Prediction of Heat Generation in a Porous Fin from Surface Temperature” (Ranjan Das and Balaram Kundu), (International Journal of Heat and Mass Transfer)