Abstract: The digital twin paradigm integrates information obtained from sensor data, physics models, as well as operational and inspection/maintenance/repair history of a physical system or component of interest. As more and more data become available, the resulting updated model becomes increasingly accurate in predicting future behavior of the system, and can potentially be used to support several objectives, such as sustainment, mission planning, and operational maneuvers. After summarizing recent developments, the talk will focus on developing a scalable probabilistic digital twin approach to account for the uncertainty regarding system properties, operational parameters, usage and environment, as well as uncertainties in data and the prediction models. The dynamic Bayesian network (DBN) methodology is used to aggregate all the uncertainty information and build a probabilistic digital twin. The DBN supports uncertainty quantification in both diagnosis and prognosis (considering both aleatory and epistemic uncertainty sources), and decision-making for system resilience. In particular, we address the challenge of scaling up the DBN-based probabilistic digital twin methodology to support real-time decision-making under uncertainty. Several strategies that combine recent advances in sensing, computing, data fusion and machine learning are developed to enable the scale-up. The digital twin computations are speeded up by leveraging sensitivity analysis, machine learning, dimension reduction, and also techniques specific to the Bayesian network, such as arc reversal and network collapsing. The developed computational techniques for scaling up probabilistic digital twins are demonstrated for several use cases related to aircraft, rotorcraft, marine vessels, and additive manufacturing. Several types of resilience-related decisions are considered, such as system sustainment (inspection scheduling, predictive maintenance), mission planning, flight maneuver adaptation, and manufacturing process quality control.

Bio: Professor Sankaran Mahadevan (Vanderbilt University, Nashville, TN, USA) has more than thirty-five years of research and teaching experience in uncertainty quantification, risk and reliability analysis, machine learning, mechanical systems diagnosis and prognosis, and decision-making under uncertainty. He has applied these methods to a variety of structures, materials and systems in civil, mechanical and aerospace engineering. His research has been extensively funded by NSF, NASA, DOE, DOD, FAA, NIST, as well as GM, Chrysler, GE, Union Pacific, and Mitsubishi, and he has co-authored two textbooks and more than 350 peer-reviewed journal papers. During the past decade, he has been at the forefront of academic research on digital twin methodologies for aircraft, rotorcraft, ship structures, and additive manufacturing, funded by FAA, U.S. Air Force, U. S. Army, and NIST. Professor Mahadevan has served as General Chair of several prominent conferences such as the AIAA SDM Conference, AIAA Non-Deterministic Approaches Conference, ASCE Engineering Mechanics Conference, and three PHM Society Annual Conferences, including PHM 2022. He is a Distinguished Member of ASCE, and Fellow of three other organizations: AIAA, Engineering Mechanics Institute, and PHM Society. He is a winner of the Alfredo Ang Award for Risk Management of Civil Infrastructure from ASCE, and the Distinguished Research Award from the International Association of Structural Safety and Reliability.

Abstract: Because of its industrial and scientific relevance, the fault diagnosis problem has been approached in many communities. In model-based diagnosis, mathematical models derived from physical insights while data-driven fault diagnosis use historical data. While model-based diagnosis is based on a rigorous theoretical framework, data-driven fault diagnosis is often treated as a general classification problem. The potential of data-driven models to capture complex relationships in data is complicated by lack of representative training data. This presentation will focus on how to combine data-driven models with theory and methods from model-based diagnosis to reason about unknown faults. Both industrial and academic examples are used to illustrate how to systematically design data-driven models using physical insights and how to apply these models in a consistency-based diagnosis framework.

Bio: Daniel Jung is an Associate Professor at the Dept. of E.E. at Linköping University, Sweden, and is part of the Swedish Excellence Center in Information Technology (ELLIIT). Daniel finished his Ph.D. at Linköping University in 2015 and did his postdoc at Ohio State University in 2017. His current research interests include fault diagnosis, with focus on hybrid fault diagnosis, and electrification of transportation.

Abstract: ProgPy is an open-source python package supporting research and development of prognostics and health management (PHM) tools, developed by NASA and several partners in the wider PHM community. ProgPy implements common functionalities of prognostics, including tools to monitor the state of physical systems and predict how they will degrade with use. It is built with a modular architecture, allowing the built-in prognostics tools and methods to work on any model, whether one distributed with ProgPy or built by a user. This talk will provide an overview of the ProgPy package, discuss what is included in ProgPy (e.g., existing models, state estimation tools, prediction methods), detail how to understand results, and highlight various applications.

Bio: Chetan S. Kulkarni is a KBR Sr. Technical Fellow and staff researcher at NASA Ames Research Center's Prognostics Center of Excellence, specializes in Systems Diagnostics, Prognostics and Health Management. His work centers on developing physics-based models and prognostics for electronic, energy, and exploration ground systems. He holds an MS ('09) and Ph.D. ('13) from Vanderbilt University, where he was a Graduate Research Assistant and BE (’02) University of Pune. Prior to joining Vanderbilt University, he conducted research at IIT-Bombay, focusing on low-cost substation automation and high voltage transformers. Earlier in his career, he was part of Honeywell's Power Automation group in India, contributing to power automation projects and product development. He is Fellow of the PHM Society (FPHMS), AIAA Associate Fellow and SMIEEE. He serves as an Associate Editor for IEEE, SAE, and IJPHM Journals on Prognostics and Systems Health Management and has co-chaired Technical Program Committees at PHME18 and PHM 20-24.