A Digital Twin Framework for Mechanical System Health State Estimation⁎This work was supported by Dow's University Partner Initiative Program and, in part, by NSF. Issue 20 (2021)

• Record Type: Journal Article
• Authors: Toothman, Maxwell; Braun, Birgit; Bury, Scott J.; Dessauer, Michael; Henderson, Kaytlin; Phillips, Steven; Ye, Yixin; Tilbury, Dawn M.; Moyne, James; Barton, Kira
• Is Part Of: IFAC-PapersOnLine. Volume 54:Issue 20(2021)
• Page Start: 1
• Page End: 7
• Publication Date: 2021
• Subjects: Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2021.11.144
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: A framework to accurately and reliably estimate mechanical system health is essential for manufacturing plants that implement a condition-based maintenance strategy. Ideally, a plant's approach to health state estimation would be uniform across systems, making it possible to consistently identify faults and reuse modeling resources. Existing health state estimation approaches, though, typically focus on identifying the presence of a single class of faults or are specific to a single type of mechanical system. This paper presents a quantitative definition of the health state estimation problem that is general to mechanical manufacturing systems. A digital twin framework that allows multiple dimensions of system health to be modeled and estimated simultaneously is then detailed. A case study implementing this framework on an industrial pump system is provided.