• 대한음성학회지:말소리
• /
• 제58호
• /
• pp.67-81
• /
• 2006

For a large corpus of time-aligned data, HMM based approaches are most widely used for automatic segmentation, providing a consistent and accurate phone labeling scheme. There are two methods for training in HMM. Flat starting method has a property that human interference is minimized but it has low accuracy. Bootstrap method has a high accuracy, but it has a defect that manual segmentation is required In this paper, a new algorithm is proposed to minimize manual work and to improve the performance of automatic segmentation. At first phase, voiced, unvoiced and silence classification is performed for each speech data frame. At second phase, the phoneme sequence is aligned dynamically to the voiced/unvoiced/silence sequence according to the acoustic phonetic rules. Finally, using these segmented speech data as a bootstrap, phoneme model parameters based on HMM are trained. For the performance test, hand labeled ETRI speech DB was used. The experiment results showed that our algorithm achieved 10% improvement of segmentation accuracy within 20 ms tolerable error range. Especially for the unvoiced consonants, it showed 30% improvement.

• International Journal of Naval Architecture and Ocean Engineering
• /
• 제12권1호
• /
• pp.226-240
• /
• 2020

This paper presents an experimental investigation of asymmetric impact effects on hydroelastic responses. A 1:64 scaled segmented ship model with U-shape open cross-section backbone was newly designed to meet elastic similarity conditions of vertical, horizontal and torsional stiffness simultaneously. Different wave heading angles and wavelengths were adopted in regular wave test. In head wave condition, parametric rolling phenomena happened along with asymmetric slamming forces, the relationship between them was disclosed at first time. The impact forces on starboard and port sides showed alternating asymmetric periodic changes. In oblique wave condition, nonlinear springing and whipping responses were found. Since slamming phenomena occurred, high-frequency bending moments became an important part in total bending moments and whipping responses were found in small wavelength. The wavelength and head angle are varied to elucidate the relationship of springing/whipping loads and asymmetric impact. The distributions of peaks of horizontal and torsional loads show highly asymmetric property.

• Computers and Concrete
• /
• 제29권2호
• /
• pp.69-79
• /
• 2022

For long underground box utility tunnels, post-tensioned precast concrete is often used. Between precast tunnel segments, sealed waterproof flexible joints are often specified. Fault displacement can lead to excessive deformation of the joints, which can lead to reduction in waterproofing due to diminished contact pressure between the sealant strip and the tunnel segment. This paper authenticates utilization of a finite element model for a prefabricated tunnel fault-crossing founded on ABAQUS software. In addition, material parameter selection, contact setting and boundary condition are reviewed. Analyzed under normal fault action are: the influence of fault displacement; buried depth; soil friction coefficient, and angle of crossing at the fault plane. In addition, distribution characteristics of the utility tunnel structure for vertical and longitudinal/horizontal relative displacement at segmented interface for the top and bottom slab are analyzed. It is found that the effect of increase in fault displacement on the splice joint deformation is significant, whereas the effects of changes in burial depth, pipe-soil friction coefficient and fault-crossing angle on the overall tunnel and joint deformations were not so significant.

• 한국측량학회지
• /
• 제30권1호
• /
• pp.1-10
• /
• 2012

최근 공간정보 기술은 정확도와 효율성 측면에서 큰 발전을 이루어 왔다. 특히, 항공 레이저 스캐너로부터 획득한 점군집 데이터를 이용하여 3차원 공간정보를 획득할 수 있게 되었다. 다양한 3차원 공간 데이터 구축에 대한 연구는 국내외의 관심 분야이며, 객체 모델링은 가장 중요한 과정이다. 본 연구의 목적은 건물 모델링의 자동화 알고리즘 개발과 이를 검증할 수 있는 시뮬레이션 데이터의 생성이다. 시뮬레이션 데이터는 건물의 다양성을 고려하여 경사형, 피라미드형, 돔형, 복합 다각형과 같은 여러 복잡한 형태의 지붕으로 구성된 객체이다. 이 논문에서는 면교차점(Model key point) 결정을 통한 자동 건물 모델링을 위하여 지붕면 패치를 기하학적 특징을 기반으로 분할하였다. 실험 결과로부터 분할된 면들은 최적의 수학적 함수에 의해 모델링 되며, 객체를 구성하는 면교차점들을 추출할 수 있었고, 인공지물에 대한 수치도화 제작을 위한3차원 도화가 가능하였다.

• 대한의용생체공학회:의공학회지
• /
• 제44권2호
• /
• pp.139-146
• /
• 2023

Osteoporosis is a disease in which the risk of bone fractures increases due to a decrease in bone density caused by aging. Osteoporosis is diagnosed by measuring bone density in the total hip, femoral neck, and lumbar spine. To accurately measure bone density in the lumbar spine, the vertebral region must be segmented from the lumbar X-ray image. Deep learning-based automatic spinal segmentation methods can provide fast and precise information about the vertebral region. In this study, we used 695 lumbar spine images as training and test datasets for a deep learning segmentation model. We proposed a lumbar automatic segmentation model, CM-Net, which combines the center point of the spine and the modified U-Net network. As a result, the average Dice Similarity Coefficient(DSC) was 0.974, precision was 0.916, recall was 0.906, accuracy was 0.998, and Area under the Precision-Recall Curve (AUPRC) was 0.912. This study demonstrates a high-performance automatic segmentation model for lumbar X-ray images, which overcomes noise such as spinal fractures and implants. Furthermore, we can perform accurate measurement of bone density on lumbar X-ray images using an automatic segmentation methodology for the spine, which can prevent the risk of compression fractures at an early stage and improve the accuracy and efficiency of osteoporosis diagnosis.

• Smart Structures and Systems
• /
• 제16권1호
• /
• pp.163-181
• /
• 2015

This research investigates the structural health monitoring of nonlinear structures after a major seismic event. It considers the identification of flag-shaped or pinched hysteresis behavior in response to structures as a more general case of a normal hysteresis curve without pinching. The method is based on the overall least squares methods and the log likelihood ratio test. In particular, the structural response is divided into different loading and unloading sub-half cycles. The overall least squares analysis is first implemented to obtain the minimum residual mean square estimates of structural parameters for each sub-half cycle with the number of segments assumed. The log likelihood ratio test is used to assess the likelihood of these nonlinear segments being true representations in the presence of noise and model error. The resulting regression coefficients for identified segmented regression models are finally used to obtain stiffness, yielding deformation and energy dissipation parameters. The performance of the method is illustrated using a single degree of freedom system and a suite of 20 earthquake records. RMS noise of 5%, 10%, 15% and 20% is added to the response data to assess the robustness of the identification routine. The proposed method is computationally efficient and accurate in identifying the damage parameters within 10% average of the known values even with 20% added noise. The method requires no user input and could thus be automated and performed in real-time for each sub-half cycle, with results available effectively immediately after an event as well as during an event, if required.

• 한국음향학회지
• /
• 제16권5호
• /
• pp.50-59
• /
• 1997

본 논문에서는 한국어 음성 데이터베이스 구축을 위하여 자동으로 음소경계를 추출하는 자동 음성분할 및 레이블링 시스템을 구현하였다. 기존의 음성분할 및 레이블링 기술을 근간으로 본 시스템을 구현하였으며, 또한 사용자가 자동분할된 음소경계를 확인하여 그 경계를 쉽게 수정할 수 있도록 한글 모티프 환경에서 그래픽 사용자 인터페이스를 개발하였다. 개발된 시스템은 16kHz로 샘플링된 음성을 대상으로 하고 있으며, 레이블링 단위는 45개의 유사음소와 하나의 묵음으로 구성하였다. 그리고 언어학적 정보의 입력방식으로는 음소표기와 철자표기를 사용하였으며, 패턴매칭 방법으로는 hidden Markov model(HMM)을 이용하였다. 개발된 시스템의 각 음소 모델은 수작업에 의해서 음소단위로 분할한 음성학적으로 균형잡힌 445 단어 데이터베이스를 이용해서 훈련되었다. 그리고 본 시스템의 성능평가를 위해 훈련에 사용되지 않는 문장 데이터베이스에 대해서 자동 음성분할 실험을 수행하였다. 실험결과, 수작업에 의해서 분할된 음소경계위치와의 오차가 20ms 이내인 것이 74.7%였으며, 40ms이내에는 92.8%가 포함되었다.

• International Journal of Naval Architecture and Ocean Engineering
• /
• 제7권5호
• /
• pp.873-887
• /
• 2015

The springing response of a large blunt ship was considered to be influenced by a second order interaction between the incoming irregular wave and the blunt geometry of the forebody of the ship. Little efforts have been made to simulate this complicated fluid-structure interaction phenomenon under irregular waves considering the second order effect; hence, the above mentioned premise still remains unproven. In this paper, efforts were made to quantify the second order effect between the wave and vibrating flexible ship structure by analyzing the experimental data obtained through the model basin test of the scaled-segmented model of a large blunt ship. To achieve this goal, the measured vertical bending moment and the wave elevation time history were analyzed using a higher order spectral analysis technique, where the quadratic interaction between the excitation and response was captured by the cross bispectrum of two randomly oscillating variables. The nonlinear response of the vibrating hull was expressed in terms of a quadratic Volterra series assuming that the wave excitation is Gaussian. The Volterra series was then orthogonalized using Barrett's procedure to remove the interference between the kernels of different orders. Both the linear and quadratic transfer functions of the given system were then derived based on a Fourier transform of the orthogonalized Volterra series. Finally, the response was decomposed into a linear and quadratic part to determine the contribution of the second order effect using the obtained linear and quadratic transfer functions of the system, combined with the given wave spectrum used in the experiment. The contribution of the second order effect on the springing response of the analyzed ship was almost comparable to the linear one in terms of its peak power near the resonance frequency.

• Journal of Animal Science and Technology
• /
• 제65권3호
• /
• pp.588-595
• /
• 2023

The aim of this study was to investigate the effects of heat stress on milk traits in South Korea using comprehensive data (dairy production and climate). The dataset for this study comprised 1,498,232 test-day records for milk yield, fat- and protein-corrected milk, fat yield, protein yield, milk urea nitrogen (MUN), and somatic cell score (SCS) from 215,276 Holstein cows (primiparous: n = 122,087; multiparous: n = 93,189) in 2,419 South Korean dairy herds. Data were collected from July 2017 to April 2020 through the Dairy Cattle Improvement Program, and merged with meteorological data from 600 automatic weather stations through the Korea Meteorological Administration. The segmented regression model was used to estimate the effects of the temperature-humidity index (THI) on milk traits and elucidate the break point (BP) of the THI. To acquire the least-squares mean of milk traits, the generalized linear model was applied using fixed effects (region, calving year, calving month, parity, days in milk, and THI). For all parameters, the BP of THI was observed; in particular, milk production parameters dramatically decreased after a specific BP of THI (p < 0.05). In contrast, MUN and SCS drastically increased when THI exceeded BP in all cows (p < 0.05) and primiparous cows (p < 0.05), respectively. Dairy cows in South Korea exhibited negative effects on milk traits (decrease in milk performance, increase in MUN, and SCS) when the THI exceeded 70; therefore, detailed feeding management is required to prevent heat stress in dairy cows.

• 산업경영시스템학회지
• /
• 제40권2호
• /
• pp.78-91
• /
• 2017

In this study, we conducted a comparative study on user's perception and behavior on public system service (PSS) using institutionalism theory and MGA (multi-group analysis) methodology. In particular, this study focuses on how institutional isomorphism is applied to public system services and how MGA can be implemented correctly in a variance based SEM (structural equation model) such as PLS (partial least square). A data set of 496 effective responses was collected from pubic system users and an empirical research was conducted using three segmented models categorized by public proximity theory (public firms = 113, government contractors = 210, private contractors = 173). For rigorous group comparisons, each model was estimated by the same indicators and approaches. PLS-SEM was used in testing research hypotheses, followed by parametric and non-parametric PLS-MGA procedures in testing categorical moderation effects. This study applied novel procedures for testing composite measurement invariance prior to multi-group comparisons. The following main results and implications are drawn : 1) Partial measurement invariance was established. Multi-group analysis can be done by decomposed models although data can not be pooled for one integrated model. 2) Multi-group analysis using various approaches showed that proximity to public sphere moderated some hypothesized paths from quality dimensions to user satisfaction, which means that categorical moderating effects were partially supported. 3) Careful attention should be given to the selection of statistical test methods and the interpretation of the results of multi-group analysis, taking into account the different outcomes of the PLS-MGA test methods and the low statistical power of the moderating effect. It is necessary to use various methods such as comparing the difference in the path coefficient significance and the significance of the path coefficient difference between the groups. 4) Substantial differences in the perceptions and behaviors of PSS users existed according to proximity to public sphere, including the significance of path coefficients, mediation and categorical moderation effects. 5) The paper also provides detailed analysis and implication from a new institutional perspective. This study using a novel and appropriate methodology for performing group comparisons would be useful for researchers interested in comparative studies employing institutionalism theory and PLS-SEM multi-group analysis technique.