Abstract: Artificial Intelligence (AI) is being explored in almost every engineering discipline in recent years. As a purely data-driven approach, classical AI is known to have poor prediction capability since no physics knowledge is included. This talk presents the concept of Hybrid Intelligence, a physics-guided learning approach that integrates machine learning (artificial intelligence) and physics (human intelligence) for engineering analysis across very different scales in aerospace. The physics-guided learning methodology encodes the underlying physics of engineering problems into machine learning models, and fuses information from abstracted knowledge and observed data together. Three different physics-encoding methods are presented to cover most common engineering problems: encoding in network architecture, encoding in input features, and encoding in output functions. The first method is illustrated using convolutional neural networks (CNN)-based method for partial differential equations (PDEs) and recurrent neural networks (RNN)-based method for ordinary differential equations (ODEs). The second method is illustrated encoding prior physics and knowledge in imaging quantifies to facilitate the learning process. The last method is illustrated using Bayesian Entropy Network (BEN)-based method for constrained probabilistic inference and uncertainty quantification. These different methods are demonstrated with aerospace engineering applications at very different scales, including material fatigue and fracture, structural dynamics, and air traffic control and management. The results show several major advantages of physics-guided learning: superior prediction/extrapolation capability; significant reduction of training sample requirements; fast computation especially for nonlinear and dynamic problems; and learning results are interpretable with physics. The limitation is that the analysts need to have certain level of domain knowledge and a pure data scientist may not be well suited for this type of learning task. Several future research directions are discussed based on the findings from the current study.

 

Bio: Yongming Liu is a Professor in mechanical and aerospace engineering at Arizona State University and is the founding director of the Center for Complex System Safety – a joint center of Arizona State University, University of Arizona, and Northern Arizona University. He completed his PhD at Vanderbilt University in 2006, and obtained his Bachelors’ and Masters’ degrees from Tongji University in China in 1999 and 2002, respectively. Dr. Liu’s research interests include prognostics, probabilistic methods, fatigue and fracture, imaging-based experiments, and Bayesian methods. He is the recipient of the Air Force Young Investigator Award in 2011 and has worked with a wide variety of governmental agencies and industrial partners for his research, including NASA, NSF, DOE, DOT, and DOD.  He is currently leading a NASA University Leadership Initiative project on information fusion for air traffic management, which focuses on the development of probabilistic learning methods with extensive data resources for air transportation safety assurance.


Physics-Guided Learning in Aerospace: From Material Fracture to System Operation

Division / Sponsors: 
Aerospace, Bioengineering, Applied Mechanics, Design Engineering Division, Advanced Energy Systems, Manufacturing Engineering Division, Noise Control and Acoustics, Nondestructive Evaluation, Diagnosis, and Prognosis (NDPD), Nuclear Engineering Division, Process Industries Division, Fluids Engineering, Heat Transfer, Materials, Microelectromechanical Systems, Safety Engineering and Risk Analysis, Computers & Information in Engineering


Paper Proceedings: https://asmedigitalcollection.asme.org/IMECE/IMECE2020/volume/84478

IMECE2020 - Virtual Conference

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.