 | "Anomaly detection in event-based manufacturing systems using model generation"Dawn Tilbury Mechanical Engineering Department |
Abstract
Discrete manufacturing systems commonly produce streams of events – parts arrive, machines start and finish, etc. With dozens of concurrent processes, these events do not have a well-defined order. Faults and other anomalies may manifest themselves in these data streams, but not be visible to even an expert observer.
This presentation describes a method for detecting anomalies in streams of event data, for systems which do not have a pre-defined formal model. Commonly-available information about the system is required as input to the method (e.g., which events are associated with which processes and resources). Since it is not known whether a formal model exists that can accurately characterize the manufacturing system, the method builds a set of models, thereby allowing uncertainty about the “true” behavior of the system to persist through the anomaly detection process. The performance of each model in the set is evaluated based on known “good” and “bad” streams of events; new streams are scored using a weighted average of the individual model’s scores, based on whether each model accepts the new stream.
The method has been applied to a Ford machining line to find an anomaly associated with a gantry incorrectly waiting for a machine to become available.
Biography
Dawn M. Tilbury received the B.S. degree in Electrical Engineering, summa cum laude, from the University of Minnesota in 1989, and the M.S. and Ph.D. degrees in Electrical Engineering and Computer Sciences from the University of California, Berkeley, in 1992 and 1994, respectively. In 1995, she joined the Mechanical Engineering Department at the University of Michigan, Ann Arbor, where she is currently a Professor. She won the EDUCOM Medal (jointly with Professor William Messner of Carnegie Mellon University) in 1997 for her work on the web-based Control Tutorials for Matlab. She is co-author (with Joseph Hellerstein, Yixin Diao, and Sujay Parekh) of the textbook Feedback Control of Computing Systems. She received an NSF CAREER award in 1999, and is the 2001 recipient of the Donald P. Eckman Award of the American Automatic Control Council. Her research interests include distributed control of mechanical systems with network communication, logic control of manufacturing systems, and dynamic modeling and control of physiological systems. She belongs to ASME, IEEE, and SWE. She is currently Vice-Chair of the ASME Dynamic Systems and Control Division Executive Committee, and will be Program Chair of the 2012 American Control Conference (ACC) and General Chair of the 2014 ACC.