Guest Faculty Research Participant - Peng Chen

Argonne National Laboratory • Lemont, IL

3d

About The Position

The proposed research focuses on advancing scientific machine learning techniques for geoscience applications, particularly in flood prediction by leveraging physics-informed machine learning methods. We will explore two innovative architectures based on our recent developments: a latent attention neural operator, which efficiently captures temporal correlations in complex dynamical systems, and the latent dynamics network, which integrates conditional physics-informed neural networks (PINNs) to enhance accuracy and generalizability. These methodologies have demonstrated success in modeling complex dynamical systems including nonlinear diffusion reaction equations, Navier—Stokes equations, and shallow water equations, and will be further adapted to improve predictive capabilities for flood events in the Midwest region, covering northern Illinois, southern Wisconsin, and western Iowa. The study will use shallow water equations as the physical model and leverage physics-based flood simulation results as training data, with the goal of developing a efficient, robust, physics-informed, and data-driven model for flood prediction in supporting energy utility planning. This research aims to ridge physics-based modeling with machine learning approaches, addressing practical challenges in real-world geoscience applications.

What This Job Offers

• Job Type: Part-time
• Education Level: No Education Listed
• Number of Employees: 1,001-5,000 employees