Advance Generalizability of Graph-based Machine Learning Models for Applications Automotive Metal Forming and Impact

In modern automotive engineering, vehicles are primarily designed in the virtual space to enable a rapid vehicle design process. However, this process is heavily constrained by the time and computational requirements necessary to generate the vast number of simulations needed for vehicle design. Fortunately, modern machine learning (ML) techniques may be used to dramatically accelerate the generation of new simulation results. In this project, several recently developed ML frameworks will be applied to industrially applicable metal forming and impact problems to speed up the vehicle design process. The ML models developed will maintain high accuracy in key performance indicators over a range of geometries, material parameters, and process parameters. The technology developed in this project will underpin the simulation framework that Impact AI is developing to provide its customers with the ability to significantly accelerate impact and stamping simulations as used in the automotive industry.

Faculty Supervisor:

Cliff Butcher;Kaan Inal

Student:

Partner:

Impact AI

Discipline:

Engineering

Sector:

Information and cultural industries

University:

University of Waterloo

Program: