Events Calendar

Astani Civil and Environmental Engineering Seminar

Speaker: Yujie Ying, Ph.D. Candidate, Department of Civil and Environmental Engineering Carnegie Mellon University

Talk Title: Data-Driven Ultrasonics for Pipe Monitoring

Abstract: Structural Health Monitoring (SHM) is the continuous assessment of structural integrity through permanently installed sensors. SHM complements traditional nondestructive evaluation (NDE) techniques that are time and labor intensive and hence infrequent. SHM allows condition-based structural maintenance to replace the current practice of economically inefficient schedule-based maintenance. However, a major challenge for SHM lies in distinguishing the damage induced changes in the sensed signal from the changes produced by benign environmental and operational variations (such as temperature, air pressure in pipes, fluid flow, etc.).

In this talk, I will present a data-driven methodology for infrastructure monitoring that is robust to ambient environmental variability. I will focus on demonstrating the effectiveness of an integrated machine learning and signal processing approach for damage detection in pipe structures with surface-mounted piezoelectric wafers that generate and sense ultrasonic waves. Ultrasonic waves can propagate long distances with high sensitivity to damage. However, it is difficult to recognize a defect due to the presence of multiple dispersive wave modes and the influence of environmental and operational factors. Laboratory experiments and field tests were conducted on a pipe specimen with randomly-controlled internal air pressure levels, and on an operating hot-water pipe with large and uncontrollable environmental fluctuations, respectively. The sensed ultrasonic data are characterized and mapped onto a high dimensional feature space using various signal processing techniques. Machine learning algorithms are then applied to automatically identify effective features and to detect a weak scatterer on the pipes under complex and highly dynamic operating conditions.

This data-driven monitoring methodology involves an integrated process of sensing, data acquisition, signal analysis, and pattern recognition, for continuous tracking of the structural functionality in an adaptive and cost-effective manner. The techniques developed in this work are expected to have broader applications related to the regular inspection, maintenance, and management of critical infrastructures not limited to pipes.


Host: Astani CEE Department

Location: Kaprielian Hall (KAP) - 209

Audiences: Everyone Is Invited

Contact: Cassie Cremeans