Friction Stir Welding, Reduced Order Models and Neural Networks

Speaker:

Huaxiong Huang, professor of applied mathematics, York University and Beijing Normal University

Description:

Friction stir welding is a preferred method for joining pieces of metal in the manufacturing processes. In order to find optimal operating conditions, computational models are used to complement experiments. However, searching over high-dimensional parameter space is computationally expensive. In this talk, we present a combined machine learning and model reduction technique to solve a 2D model for friction stir welding that consists of the Navier-Stokes equations with a heat equation. Numerical results will be given to illustrate the significant reduction of computational time and effectiveness of the method.

This is joint work with my former postdocs X. Cao, Z. Song, in collaboration with the National Research Council of Canada.

Multiscale Modeling and Computation