PERL: Probabilistic energy-ratio-based localization for boiler tube leaks using descriptors of acoustic emission signals. (February 2023)

• Record Type: Journal Article
• Authors: Na, Kyumin; Yoon, Heonjun; Kim, Jaedong; Kim, Sungjong; Youn, Byeng D.
• Is Part Of: Reliability engineering & system safety. Volume 230(2023)
• Publication Date: 2023-02
• Subjects: Probabilistic modeling -- Boiler tube -- Acoustic emission signal -- Leak localization -- Energy descriptor; Reliability (Engineering) -- Periodicals; System safety -- Periodicals; Industrial safety -- Periodicals; Fiabilité -- Périodiques; Sécurité des systèmes -- Périodiques; Sécurité du travail -- Périodiques; 620.00452
• DOI: 10.1016/j.ress.2022.108923
• Languages: English
• ISSNs: 0951-8320

Abstract

Highlights: Probabilistic energy-ratio-based localization (PERL) method is proposed. PERL combines the acoustic dissipation model with sensor uncertainty model. PERL can robustly estimate the source of leak signal via time sequential updating. The effectiveness of PERL for boiler tube leak is validated via real industrial data. Abstract: This paper proposes a novel method for boiler tube leak localization in a thermal power plant, using acoustic emission sensors. In industrial settings, due to computational and storage capacity, the measured acoustic emission signal is often processed through the use of descriptors, such as the root mean square (RMS), which is related to the signal energy. Computational and storage capacity issues make it difficult to use conventional methods, including time difference of arrival, which uses a high-sampling-rate signal. In addition, the measured RMS may have uncertainty that arises due to sensor disturbance or unpredictable process conditions. Thus, this study newly proposes an approach called probabilistic energy-ratio-based localization (PERL) to estimate the location of a boiler tube leak. In the proposed approach, acoustic dissipation theory is used to calculate the ratio of the signal energy from the specific band energy. To account for background noises and sensor disturbance, the uncertainty of the measured RMS is characterized in a probabilistic manner. Using this information, the probability that a boiler tube leak has occurred at a specific location is estimated hypothetically. Case studies confirm that the proposed method enables localization of a boiler tube leak position with high accuracy.