2024 Undergraduate Research Showcase

Harnessing the Power of AI: Revolutionizing Engineering with Physics-Informed Neural Networks

Document Type

Student Presentation

Presentation Date

4-19-2024

Faculty Sponsor

Dr. Mahmood Mamivand

Abstract

Since May 2023, my time as a Student Research Technician at Boise State University, has been pivotal in transforming the landscape of computational physics through the development and optimization of Physics-Informed Neural Networks (PINNs). By leveraging Python, TensorFlow, and DeepXDE, I have focused on modeling complex partial differential equations, particularly those with challenging boundary conditions. This innovative approach has led to substantial improvements in model convergence, achieving reductions in loss that far exceed the capabilities of traditional computational methods.

The employment of advanced deep learning techniques, coupled with the strategic utilization of Boise State's Borah supercomputer cluster, has significantly enhanced the project's efficiency and scalability. My contributions have not only showcased the potential of integrating machine learning and artificial intelligence into scientific research but have also highlighted my skills in problem-solving, teamwork, and technical communication. This work represents a milestone in the application of AI technologies to solve complex scientific challenges, underscoring the transformative potential of PINNs in advancing computational physics and setting new benchmarks for scientific modeling.

This document is currently not available here.

Share

COinS