• ETRI Journal
• /
• v.41 no.2
• /
• pp.235-241
• /
• 2019

In this paper, we propose a rank-weighted reconstruction feature to improve the robustness of a feed-forward deep neural network (FFDNN)-based acoustic model. In the FFDNN-based acoustic model, an input feature is constructed by vectorizing a submatrix that is created by slicing the feature vectors of frames within a context window. In this type of feature construction, the appropriate context window size is important because it determines the amount of trivial or discriminative information, such as redundancy, or temporal context of the input features. However, we ascertained whether a single parameter is sufficiently able to control the quantity of information. Therefore, we investigated the input feature construction from the perspectives of rank and nullity, and proposed a rank-weighted reconstruction feature herein, that allows for the retention of speech information components and the reduction in trivial components. The proposed method was evaluated in the TIMIT phone recognition and Wall Street Journal (WSJ) domains. The proposed method reduced the phone error rate of the TIMIT domain from 18.4% to 18.0%, and the word error rate of the WSJ domain from 4.70% to 4.43%.

• Journal of Nuclear Fuel Cycle and Waste Technology(JNFCWT)
• /
• v.19 no.1
• /
• pp.75-85
• /
• 2021

Evaluating the quantitative damage to rocks through acoustic emission (AE) has become a research focus. Most studies mainly used one or two AE parameters to evaluate the degree of damage, but several AE parameters have been rarely used. In this study, several data-driven models were employed to reflect the combined features of AE parameters. Through uniaxial compression tests, we obtained mechanical and AE-signal data for five granite specimens. The maximum amplitude, hits, counts, rise time, absolute energy, and initiation frequency expressed as the cumulative value were selected as input parameters. The result showed that gradient boosting (GB) was the best model among the support vector regression methods. When GB was applied to the testing data, the root-mean-square error and R between the predicted and actual values were 0.96 and 0.077, respectively. A parameter analysis was performed to capture the parameter significance. The result showed that cumulative absolute energy was the main parameter for damage prediction. Thus, AE has practical applicability in predicting rock damage without conducting mechanical tests. Based on the results, this study will be useful for monitoring the near-field rock mass of nuclear waste repository.

• Journal of Apiculture
• /
• v.34 no.1
• /
• pp.7-13
• /
• 2019

Vespid wasps (Vespa spp.) are the most noxious pests on apiculture, resulting in significant economic losses. Early monitoring and management are the first step to prevent the damages from vespid wasps. In this study, the acoustic signals from wasps and honey bees were measured by a microphone with a preamplifier and an analog-digital converter. In frequency analysis of the acoustic signals from wasps and honey bees, there were differences between the two species. While the fundamental frequency of the wasps was analyzed to be about 100 Hz with the strong harmonic frequencies, that of the honey bees was about 200~250 Hz. The 2nd harmonic signals from wasp were strongest while the fundamental ones from honey bees were. These different sound features generated by wasps or honey bees might be applied to develop the early monitoring system of the incursion of wasps to the apiary.

• Journal of the Korean Society of Laryngology, Phoniatrics and Logopedics
• /
• v.25 no.1
• /
• pp.20-23
• /
• 2014

Age-related changes in larynx can have a direct impact on voice quality and general comfort level. Observations of vocal aging have spanned perceptual, acoustic, aerodynamic, physical, electromyographic (EMG) and histological levels. Evidence of differential vocal aging in relation to gender and physical condition has been reported. Perceptual, acoustic, aerodynamic, kinematic, EMG and histological data document age-related changes in laryngeal structure and function with advancing age. These changes contribute to a functional age-related impact of vocal hypofunction or compensatory hyperfunction. This review will focus on the current understanding of the clinical and cellular changes in the larynx that lead to presbyphonia.

• Journal of the Korean Society for Precision Engineering
• /
• v.22 no.4
• /
• pp.37-43
• /
• 2005

Real time monitoring of exit burr formation is critical in manufacturing automation. In this paper, acoustic emission (AE) was used to detect the burr formation during drilling. By using wavelet transform (WT), AE data were compressed without unnecessary details. Then the transformed data were used as selected features (inputs) of a back-propagation artificial neural net (ANN). In order to validate the in process AE monitoring system, both WT-based ANN and cutting condition (cutting speed, feed, drill diameter, etc.) based ANN outputs were compared with experimental data.

• Proceedings of the KIEE Conference
• /
• 2004.11c
• /
• pp.49-51
• /
• 2004

This paper presents a realization of fault detection algorithm and Fieldbus based communication for condition monitoring of check valve. We first acquired the AE(Acoustic Emission) sensor data at the KAERI check valve test loop, extract fault features through the learned Neural network, and send the processed data to a remote site. The overall system has been implemented and experimental results are given to show its effectiveness.

• Journal of Ocean Engineering and Technology
• /
• v.15 no.4
• /
• pp.60-65
• /
• 2001

Thermal spray coating is formed by a process in which melted particles flying with high speed towards substrate, then crash and spread on the substrate surface cooled and solidified in a very short time, Stacking of the particles makes coating. In this study, the exfoliation of $Al_2$O$_3$ and Ni-4.5wt.%Al thermally sprayed coating which were deposited by an atmospheric plasma spray apparatus are investigated using an AE method. A tensile test is conducted on notch specimens in a stress range below the elastic limit of substrate. The wave forms of AE generated from the three coating specimens can be classified by FFT analysis into two types which low frequency(type I waveform is considered to corresponds exfoliation of coating layers and type II waveform corresponds the plastic deformation of notch tip or the resultant fracture of coating. The fracture of the coating layers can estimate by AE event and amplitude, because AE features increase when the deformation generates.

• Journal of information and communication convergence engineering
• /
• v.9 no.5
• /
• pp.510-514
• /
• 2011

To detect the insulator in the fault state on the electric poles, we first measured radiation sounds from normal state insulators and error state insulators in the anechoic chamber. We processed the signals in frequency domain to find the features with filter bank, narrow band and wide band analysis. So we could found two apparent results from their frequency spectrums - one was 120Hz harmonic components, the other was high average noise level than normal state ones. Then we also introduced a technique for the direction detection of the fault state insulator using the cross correlation from the three dimensional array microphones. To eliminate the noise signal from unexpected directions, we suggested the zero padding technique in cross correlation function. From these, we could conclude that acoustic fault detection techniques are useful of the detection of insulators' faults and the estimation of the direction of the fault state insulators.

• Journal of Mechanical Science and Technology
• /
• v.17 no.11
• /
• pp.1599-1607
• /
• 2003

The characteristics of elastic waves emanated from crack initiation in 6061 aluminum alloy subjected to fatigue loading are investigated through experiments. The objective of the study is to determine the differences in the properties of the signals generated from fatigue test and also to examine if the sources of the waves could be identified from the temporal and spectral characteristics of the acoustic emission (AE) waveforms. The signals are recorded using nonresonant, flat, broadband transducers attached to the surface of the alloy specimens. The time dependence and power spectra of the signals recorded during the tests were examined and classified according to their special features. Six distinct types of signals were observed. The waveforms and their power spectra were found to be dependent on the crack propagation stage and the type of fracture associated with the signals. The potential application of the approach in health monitoring of structural components using a network of surface mounted broadband sensors is discussed.

• Korean Journal of Fisheries and Aquatic Sciences
• /
• v.50 no.2
• /
• pp.195-206
• /
• 2017

Joint time-frequency images of the broadband echo signals of six fish species were obtained using the smoothed pseudo-Wigner-Ville distribution in controlled environments. Affine transformation and principal component analysis were used to obtain eigenimages that provided species-specific acoustic features for each of the six fish species. The echo images of an unknown fish species, acquired in real time and in a fully automated fashion, were identified by finding the smallest Euclidean or Mahalanobis distance between each combination of weight matrices of the test image of the fish species to be identified and of the eigenimage classes of each of six fish species in the training set. The experimental results showed that the Mahalanobis classifier performed better than the Euclidean classifier in identifying both single- and mixed-species groups of all species assessed.