Please join us Wednesday May 6 at 1pm ET/11am MT for the next installment of our Georgia Tech - Sandia virtual seminar series featuring Dr. Keegan Moore, Associate Professor in the Daniel Guggenheim School of Aerospace Engineering at the Georgia Institute of Technology, who will be presenting on Data-Driven Identification and Reduced-Order Modeling of Strongly Nonlinear Structures
An EQ-4B Global Hawk, one of the United States’ most advanced and expensive unmanned aerial vehicles, lost stability and crashed in 2011 because a single screw loosened. Failures such as this reflect a broader challenge in engineering: we still lack practical tools for identifying, modeling, and predicting the nonlinear dynamics that govern real structures. This talk presents our recent research on this challenge through two complementary thrusts: physics-informed model discovery and reduced-order modeling. The first thrust focuses on identifying governing equations and parameters directly from data. This includes the Energy-Based Dual-Phase Dynamics Identification (EDDI) method for constructing equations of motion for nonlinear oscillators and the System Identification via Validation and Adaptation (SIVA) framework for simultaneous identification, validation, and uncertainty quantification. The second thrust focuses on reduced-order models for evolving structu ral interfaces, with particular emphasis on bolted-joint loosening, where bolt tension is treated as a dynamic degree of freedom that governs the effective stiffness and damping of the joint. Together, these efforts illustrate how local interface evolution can produce complex global dynamics, and how combining physics, data, and experiments can yield predictive and interpretable models for nonlinear structures.