• Geophysics and Geophysical Exploration
• /
• v.25 no.4
• /
• pp.177-188
• /
• 2022

Recently, the application of distributed acoustic sensors (DAS), which can replace geophones and seismometers, has significantly increased along with interest in micro-seismic monitoring technique, which is one of the CO₂ storage monitoring techniques. A significant amount of temporally and spatially continuous data is recorded in a DAS monitoring system, thereby necessitating fast and accurate data processing techniques. Because event detection and seismic phase picking are the most basic data processing techniques, they should be performed on all data. In this study, a machine learning-based P, S wave phase picking algorithm was developed to compensate for the limitations of conventional phase picking algorithms, and it was modified using a transfer learning technique for the application of DAS data consisting of a single component with a low signal-to-noise ratio. Our model was constructed by modifying the convolution-based EQTransformer, which performs well in phase picking, to the ResUNet structure. Not only the global earthquake dataset, STEAD but also the augmented dataset was used as training datasets to enhance the prediction performance on the unseen characteristics of the target dataset. The performance of the developed algorithm was verified using K-net and KiK-net data with characteristics different from the training data. Additionally, after modifying the trained model to suit DAS data using the transfer learning technique, the performance was verified by applying it to the DAS field data measured in the Pohang Janggi basin.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.2
• /
• pp.139-147
• /
• 2021

This paper introduces and implements a Sound Event Detection (SED) system based on weakly-supervised learning where only part of the data is labeled, and analyzes the effect of parameters. The SED system estimates the classes and onset/offset times of events in the acoustic signal. In order to train the model, all information on the event class and onset/offset times must be provided. Unfortunately, the onset/offset times are hard to be labeled exactly. Therefore, in the weakly-supervised task, the SED model is trained by "strongly labeled data" including the event class and activations, "weakly labeled data" including the event class, and "unlabeled data" without any label. Recently, the SED systems using the mean-teacher model are widely used for the task with several parameters. These parameters should be chosen carefully because they may affect the performance. In this paper, performance analysis was performed on parameters, such as the feature, moving average parameter, weight of the consistency cost function, ramp-up length, and maximum learning rate, using the data of DCASE 2020 Task 4. Effects and the optimal values of the parameters were discussed.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.34 no.5
• /
• pp.369-376
• /
• 2014

This paper presents a method that integrates passive and active-sensing techniques for the structural health monitoring of plate-like structures. Three piezoelectric transducers are deployed in a L-shape to detect and locate an impact event by measuring and processing the acoustic emission data. The same sensor arrays are used to estimate the subsequent structural damage using guided waves. Because this method does not require a prior knowledge of the structural parameters, such as the wave velocity profile in various directions, accurate results could be achieved even on anisotropic or curved plates. A series of experiments was performed on plates, including a spar-wing structure, to demonstrate the capability of the proposed method. The performance was also compared to that of traditional approaches and the superior capability of the proposed method was experimentally demonstrated.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.23 no.3
• /
• pp.220-226
• /
• 2003

This paper presents the result of an investigation on the application of the acoustic emission method to the monitoring of fatigue crack initiation, growth and track location in welded joints. Fatigue test was carried out for a typical fillet welded joint of ship structure. AE parameter such as ring down count was analyzed in time domain and crack locations were examined by source location and cluster option which is one of the functions of AE signal processor The usability of AE mettled was confirmed for the detection of the initiation and location of through crack.

• Journal of Electrical Engineering and Technology
• /
• v.8 no.6
• /
• pp.1487-1496
• /
• 2013

In this paper, a novel algorithm for PD localization in power transformers based on wavelet de-noising technique and energy criterion is proposed. Partial discharge is one of the main failures in power transformers. The localization of which could be very useful for maintenance systems. Acoustic signals due to a PD event are transient, irregular and non-repetitive. So wavelet transform is an efficient tool for this signal processing problem that gives a time-frequency demonstration. First, different wavelet based de-noising methods are analyzed. Then, a reasonable structure for threshold value determining and applying manner on signals is presented. Evaluated errors are good evidences for choices. Next, applying the elimination low energy frequency bands is discussed and developed as a de-noising method. Time differences between signals are used for PD localization. Different ways in time arrival detection are introduced and a novel approach in energy criterion method is presented. At the end, the quality of algorithm is verified through the different assays in lab.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.36 no.3
• /
• pp.181-187
• /
• 2016

A thermal power plant has a heat exchanger tube to collect and convert the heat generated from the high temperature and pressure steam to energy, but the tubes are arranged in a complex manner. In the event that a leakage occurs in any of these tubes, the high-pressure steam leaks out and may cause the neighboring tubes to rupture. This leakage can finally stop power generation, and hence there is a dire need to establish a suitable technology capable of detecting tube leaks at an early stage even before it occurs. As shown in this paper, by applying acoustic emission (AE) technology in existing boiler tube leak detection equipment (BTLD), we developed a system that detects these leakages early enough and generates an alarm at an early stage to necessitate action; the developed system works better that the existing system used to detect fine leakages. We verified the usability of the system in a 560MW-class thermal power plant boiler by conducting leak tests by simulating leakages from a variety of hole sizes (ⵁ2, ⵁ5, ⵁ10 mm). Results show that while the existing fine leakage detection system does not detect fine leakages of ⵁ2 mm and ⵁ5 mm, the newly developed system could detect leakages early enough and generate an alarm at an early stage, and it is possible to increase the signal to more than 18 dB.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2019.06a
• /
• pp.160-162
• /
• 2019

음향 이벤트 인식 기술은 오디오 신호에서 음향 이벤트를 예측하는 기술로, 최근 대용량 데이터베이스의 배포, 인식 알고리즘과 하드웨어의 발전, 관련 인식 대회 등에 힘입어 많은 연구가 이루어지고 있는 분야이다. 본 논문에서는 음향 장면 및 이벤트 인식 관련 대회인 DCASE 챌린지에 대하여 기술하고, 약한 레이블 기반의 데이터를 학습해 강한 레이블을 예측하는 DCASE 챌린지 과제 4에 대하여 설명한다. 또한 DCASE 챌린지 과제 4에 제출된 다양한 음향 이벤트 인식 알고리즘과 데이터베이스의 종류에 따른 성능을 비교하여 음향 이벤트 인식 성능을 분석한다.

• Smart Structures and Systems
• /
• v.12 no.2
• /
• pp.97-119
• /
• 2013

This paper presents numerical and experimental results on the use of guided waves for structural health monitoring (SHM) of crack growth during a fatigue test in a thick steel plate used for civil engineering application. Numerical simulation, analytical modeling, and experimental tests are used to prove that piezoelectric wafer active sensor (PWAS) can perform active SHM using guided wave pitch-catch method and passive SHM using acoustic emission (AE). AE simulation was performed with the multi-physic FEM (MP-FEM) approach. The MP-FEM approach permits that the output variables to be expressed directly in electric terms while the two-ways electromechanical conversion is done internally in the MP-FEM formulation. The AE event was simulated as a pulse of defined duration and amplitude. The electrical signal measured at a PWAS receiver was simulated. Experimental tests were performed with PWAS transducers acting as passive receivers of AE signals. An AE source was simulated using 0.5-mm pencil lead breaks. The PWAS transducers were able to pick up AE signal with good strength. Subsequently, PWAS transducers and traditional AE transducer were applied to a 12.7-mm CT specimen subjected to accelerated fatigue testing. Active sensing in pitch catch mode on the CT specimen was applied between the PWAS transducers pairs. Damage indexes were calculated and correlated with actual crack growth. The paper finishes with conclusions and suggestions for further work.

• Journal of Mechanical Science and Technology
• /
• v.17 no.10
• /
• pp.1423-1430
• /
• 2003

In this paper, we develop a real time monitoring system to detect third-party damage on natural gas pipeline. When the damage due to third-party incidents causes an immediate rupture, the developed on-line monitoring system can help reducing the sequences of event at once. Moreover, since many third-party incidents cause damage that does not lead to immediate rupture but can grow with time, the developed on-line monitoring system can execute a significant role in reducing many third-party damage incidents. Also, when the damage is given at a point on natural gas pipeline, the acoustic wave is propagated very fast about 421.3 m/s. Therefore, the data processing time should be very short in order to detect precisely the impact position. Generally, the pipeline is laid under ground or sea and the length is very long. So a wireless data communication method is recommendable and the sensing positions are limited by laid circumstance and setting cost of sensors. The calculation and monitoring software is developed by an algorithm using the propagation speed of acoustic wave and data base system based on wireless communication and DSP systems. The developed monitoring system is examined by field testing at Balan pilot plant, KOGAS being done in order to demonstrate its validity through reactive detection of third-party contact with pipelines. Furthermore, the development system was set at the practical pipelines such as an offshore pipeline between two islands Yul-Do and Youngjong-Do, and a land branch of Pyoungtaek, Korea and it has been operating in real time.

• Journal of Mechanical Science and Technology
• /
• v.18 no.7
• /
• pp.1055-1061
• /
• 2004

In this research, a vision system for detecting breakages of small-diameter taps, which are rarely detected by the indirect in-process monitoring methods such as acoustic emission, cutting torque and motor current, was developed. Two HMI (Human Machine Interface) programs to embed the developed vision system into a Siemens open architecture controller, 840D, were developed. They are placed in sub-windows of the main window of the 840D and can be activated or deactivated either by a softkey on the operating panel or the M code in the NC part program. In the event that any type of tool breakage is detected, the HMI program issues a command for an automatic tool change or sends an alarm signal to the NC kernel. An evaluation test in a high-speed tapping machine showed that the developed vision system was successful in detecting breakages of small-diameter taps up to M1.