Title: Acoustic emission Bayesian source location: Onset time challenge

Abstract

Robust identification of the most accurate observed input data among a pool of observations is key in modeling and decision making. A statistically biased observed measurement deteriorates the predictive power of a model and affects decision-making ability based on the prediction of the model. When two competing methods of measurement are available, such as methods which identify arrival times in acoustic emission (AE) signals, a principal question is whether one of the two obtained datasets, or a combination of the two, should be used later on, for example, to localize an AE source. This question becomes more important when collecting not repeatable data such as AE signals created by a propagating crack. Here this paper considers an inverse source location problem in a concrete block to address the mentioned issue, a proposed methodology which also has wider application in competitive data selection. Elastic energy released by an AE event, such as a propagating crack, is recorded by acoustic emission data acquisition system. The onset time of AE signals is often used to locate the source of the event, and its accuracy directly affects the precision of source identification. This research proposes an innovative approach to select the most probable onsetmore » time obtained from two automatic picker methods. The proposed method selects the most probable onset times, which are observed by each picker for each sensor, in a probabilistic fashion. To validate the proposed method, the most accurate onset time observed by each picker is identified by visual inspection and is compared with the one is selected by the proposed method. Finally, the dataset is used for source location identification. Results show that picked onset times determined by the proposed method generate more accurate source identification when compared with coordinates obtained using each dataset individually.« less

Authors:

Madarshahian, Ramin  ^[1];

• Search DOE PAGES for author "Madarshahian, Ramin"
• Search DOE PAGES for ORCID "0000-0003-3444-2790"

Ziehl, Paul  ^[2];

• Search DOE PAGES for author "Ziehl, Paul"
• Search DOE PAGES for ORCID "0000-0002-4783-9255"

Caicedo, Juan M. ^[2]

---

+ Show Author Affiliations

1. University of California, San Diego, CA (United States)
2. University of South Carolina, Columbia, SC (United States)

Publication Date:

Thu Jan 24 00:00:00 EST 2019

Research Org.:

Univ. of Nebraska, Lincoln, NE (United States); Univ. of California, San Diego, CA (United States)

Sponsoring Org.:

USDOE Office of Nuclear Energy (NE)

OSTI Identifier:

1613930

Alternate Identifier(s):

OSTI ID: 1547824

Grant/Contract Number:  

NE0008544

Resource Type:

Accepted Manuscript

Journal Name:

Mechanical Systems and Signal Processing

Additional Journal Information:

Journal Volume: 123; Journal Issue: C; Journal ID: ISSN 0888-3270

Publisher:

Elsevier

Country of Publication:

United States

Language:

English

Subject:

97 MATHEMATICS AND COMPUTING; acoustic emission signal processing; onset time; data selection; data cleansing; source location; Bayesian modeling