Title:
Categorization of Passive and Active Ultrasonic Stress Wave-Based Monitoring Data from Concrete Structures Using Sequential K-Subspaces
Author(s):
Thomas Schumacher
Publication:
Web Session
Volume:
Issue:
Appears on pages(s):
Keywords:
DOI:
Date:
3/28/2022
Abstract:
Combining active and passive ultrasonic stress wave-based monitoring can provide holistic information about the condition changes of concrete structures. The passive and active approach, i.e., acoustic emission and ultrasonic coda wave monitoring, respectively, capture internal fracture processes and internal changes due to minute internal variations of stress, temperature, and degradation processes. A challenge analyzing the recorded signals is to characterize them automatically according to their cause. While solutions have been proposed in the literature, they are not generalizable. Additionally, one method deployed on its own might not give the full picture of how a structure is changing over time. The presented research introduces Sequential K-subspaces, a novel unsupervised machine learning algorithm, to assist in this complex task. Preliminary results using data from laboratory experiments are used to highlight the potential of this approach to aid in distinguishing and categorizing the recorded signals.