Implementing Generative Artificial Intelligence in Computer Aided Engineering

Abstract

The use of Finite element methods for simulation of complex engineering systems is a challenging task as it requires solving a system of equations at each time step and the same goes for implicit numerical integration of nonlinear ODEs. Moreover, to accomplish good correlation of FEM simulation with test measurements, comprehensive modeling of mechanisms in the system is essential. It thus raises cost and time requirements for modeling and analysis. The intent of this research is to explore ways to lessen computational cost while simulating outcomes for an engineering problem in real time. This task can be achieved by implementing machine learning algorithms for leveraging model’s time dependent behavior to predict response of model for parametric studies. Machine Learning is a subset of artificial intelligence which is based on the fundamental of learning from records comprising of datasets of numbers, images, text, features and determining algorithms to solve a task. In this study the linear and non-linear responses of a cantilever beam element to load application are predicted using Odyssee Lunar, a machine learning tool designed for implementing machine learning interpolation algorithms with a reduced order modeling approach. The results predicted by the machine learning algorithm are very accurate and precise when validated with finite element model.