• Journal of the Korean Society for Nondestructive Testing
• /
• v.24 no.6
• /
• pp.590-596
• /
• 2004

Neural network-based signal classification systems are increasingly used in the analysis of large volumes of data obtained in NDE applications. Ultrasonic inspection methods on the other hand are commonly used in the nondestructive evaluation of welds to detect flaws. An important characteristic of ultrasonic inspection is the ability to identify the type of discontinuity that gives rise to a peculiar signal. Standard techniques rely on differences in individual A-scans to classify the signals. This paper proposes an ultrasonic signal classification technique based on the information tying in the neighboring signals. The approach is based on a 2-dimensional Fourier transform and the principal component analysis to generate a reduced dimensional feature vector for classification. Results of applying the technique to data obtained from the inspection of actual steel welds are presented.

• Journal of Ocean Engineering and Technology
• /
• v.23 no.6
• /
• pp.117-123
• /
• 2009

This paper presents a numerical simulation of guided ultrasonic waves propagating in aluminum-epoxy-aluminum adhesive plates. In particular, this study investigated the effect of the epoxy thickness on the dispersive patterns, such as the phase velocity and group velocity of guided ultrasonic waves. In addition to investigating the dispersive curves, a numerical simulation using the pulse-echo method was carried out. This simulation showed that the degree of sensitivity of the epoxy thickness is dependent on the curvature of the phase and group dispersion curves, the maximum amplitude of the received time signals, and the peak frequency of the real components of the Fourier transform. Then, the linear relations between the epoxy thickness and the received and transformed signals were constructed to estimate the epoxy thickness.

• 제어로봇시스템학회:학술대회논문집
• /
• 2001.10a
• /
• pp.89.5-89
• /
• 2001

In this paper we explain about systems developed to forecast the standing up of the dementia patients. Basic idea of our approach is to measure the respiration and body movements in the bed. Based on the data measured, forecasting of standing up action from the bed is possible to some extent. First system proposed uses an image processing technique. Data obtained about the respiration and movements are used to forecast the standing up of the patients. The second system uses ultrasonic sensors, which emits and receives ultrasonic signals. Two techniques f the ultrasonic sensors are considered. One technique is to emit ultrasonic burst signals and to detect the instance when ...

• Proceedings of the Korean Institute of IIIuminating and Electrical Installation Engineers Conference
• /
• 2000.11a
• /
• pp.160-162
• /
• 2000

This paper describes the study on the basic characteristics of discharge generated in cable splices using ultrasonic measurement method. Two kinds of defect which might be existed in a real cable splice were simulated. One is surface discharge moisture and metallic alien substances and the other is partial discharge by metallic particle. As a result, there were some differences in the characteristics of ultrasonic signals according to each defect. Therefore, it was found that is could be possible to protect the failure in cable splices in advance by analyzing the ultrasonic signals coming from the cable splice due to partial discharge.

• Journal of Electrical Engineering and Technology
• /
• v.9 no.1
• /
• pp.259-267
• /
• 2014

This paper presents a simple, consistent and reliable technique to identify detected partial discharges (PD) using an acoustic ultrasonic method. A new reliable algorithm named 'Simple Partial Discharge Identifier' (SPDI) is proposed to perform identification process of the detected ultrasonic signals of PD. Experimental works based on recommended practices were setup and the ultrasonic signals of the PD were recorded. The PD data is then employed as the reference data. The SPDI developed has been tested against commonly used models in Neural Network (NN). Results from the SPDI algorithm shows more reliable results compared to NN models results. Comparison made on the mean square error (MSE) results shows SPDI produces the desired outcome with lower MSE in 97.17% of trials. Low error of SPDI indicates a high reliability to be applied in the identification of PD.

• Korean Journal of Air-Conditioning and Refrigeration Engineering
• /
• v.27 no.3
• /
• pp.174-178
• /
• 2015

This paper describes a signal processing technique for identifying signals from defects by using an electromagnetic-ultrasonic guided waves method based on a magnetostrictive sensor that generates a torsional mode (T(0, 1)). Because this technique is based on the digital filtering, the filtered signals provide information on the relationship between the cutoff frequency of band-pass filter and the characteristic of defect signals in heat exchange tubes. To verify the performance of the technique, artificial defects with various thickness reduction ration and shape were machined in titanium tubes, and digital filtering results are reported. The results show that digital filtering provides information to the identify shape of defects and the contact condition between the tube and support ring. Therefore, the proposed technique has good potential as a tool for evaluating the integrity of heat exchange tubes.

• Journal of the Korean Society of Manufacturing Technology Engineers
• /
• v.19 no.3
• /
• pp.407-413
• /
• 2010

This paper reports on the research which analyzes acoustic signals acquired in progressive compressing, hole blanking, and burr compacting process. An acoustic sensor was set on the bed of hydraulic press. Acoustic signal is generated from progressive stamping process. First the signal acquired from the unit process; compressing, blanking or compacting, is studied by Fourier Transform and Short Time Fourier Transform. The blanking process emitted ultrasonic signal with more than 20kHz, but the compressing and compacting processes emitted acoustic signals with lower than 10kHz. The combined signals periodically acquired right after the tool grinding were then analyzed. 70-80kHz signals appeared in time-frequency domain, but not in the frequency domain, the magnitude of which was related to the tool wear. Short Time Fourier Transform made up for the Fourier Transform in analyzing the emitted signal for stamping process in the ultrasonic domain.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.13 no.3
• /
• pp.73-81
• /
• 1999

Measurements of ultrasonic signals caused by partial discharges were perlonnrl by using an ultrasonic measurement technique to diagnose the electrical treeing deterioration in an epoxy mold-type power transformer. We also examined the characteristics of tree growth, partial discharge magnitude and ultrasonic pulse number with the increase of the deterioration time. From these results, it was found that the shape of the tree was a branch-type, and the rates of the tree growth were examined when ac voltages of 16[kV] and 20[kV] were applied Ultrasonic pulse number and discharge magnitude were rapidly increased as the length of the tree grows after the middle stage of the deterioration time, and it appeared that ultrasonic pulse number was proportional to discharge magnitude. Attenuation, time-delaying and directivity characteristics of ultrasonic signals propagated into epoxy resin by using ultrasonic oscillation and receiving systems are also re[prted as a basic data of ultrasonic mesasurements in mold-type power transformer.former.

• The Journal of the Acoustical Society of Korea
• /
• v.35 no.1
• /
• pp.74-82
• /
• 2016

In this paper, a new automatic speech recognition (ASR) was proposed where ultrasonic doppler signals were used, instead of conventional speech signals. The proposed method has the advantages over the conventional speech/non-speech-based ASR including robustness against acoustic noises and user comfortability associated with usage of the non-contact sensor. In the method proposed herein, 40 kHz ultrasonic signal was radiated toward to the mouth and the reflected ultrasonic signals were then received. Frequency shift caused by the doppler effects was used to implement ASR. The proposed method employed multi-channel ultrasonic signals acquired from the various locations, which is different from the previous method where single channel ultrasonic signal was employed. The PCA(Principal Component Analysis) coefficients were used as the features of ASR in which hidden markov model (HMM) with left-right model was adopted. To verify the feasibility of the proposed ASR, the speech recognition experiment was carried out the 60 Korean isolated words obtained from the six speakers. Moreover, the experiment results showed that the overall word recognition rates were comparable with the conventional speech-based ASR methods and the performance of the proposed method was superior to the conventional signal channel ASR method. Especially, the average recognition rate of 90 % was maintained under the noise environments.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.19 no.4
• /
• pp.386-392
• /
• 2009

This paper reports on the research which investigates acoustic signals acquired in progressive compressing, hole blanking, shearing and burr compacting process. The work piece is the head pin of the electric connector, whose raw material is the preformed steel bar. An acoustic sensor was set on the bed of hydraulic press. Because the acquired signals include the dynamic characteristics generated for all the processes, it is required to investigate signal characteristics corresponding to unit process. The corresponding dynamic characteristics to the respective process were first studied by analyzing the signals respectively acquired from compressing, blanking and compacting process. The combined signals were then periodically analyzed from the grinding to the grinding in the sound frequency domain and in the ultrasonic wave. The frequency of around 9 kHz in the sound frequency domain was much correlated to the tool wear. The characteristic frequency in the acoustic emission domain between 100 kHz and 500 kHz was not only clearly observed right after tool grinding but its amplitude was also related to the wear. The frequency amplitudes of 160 kHz and 320 kHz were big enough to be classified by the noise. The noise amplitudes are getting bigger, and their energy was much bigger as coming to the next regrinding. The signal analysis was based on the real time data and its frequency spectrum by Fourier Transform. As a result, the acousto-ultrasonic signals were much related to the tool wear progression.