• 한국군사과학기술학회지
• /
• 제23권2호
• /
• pp.115-124
• /
• 2020

Due to the increasing threats of submarines from North Korea and other countries, ROK Navy should improve the detection capability of submarines. There are two ways to detect submarines : acoustic detection and non-acoustic detection. Since the acoustic-detection way has limitations in spite of its usefulness, it should have the complementary way. The non-acoustic detection is the way to detect submarines which are operating mast sets such as periscopes and snorkels by non-acoustic sensors. So, this paper proposes a new submarine non-acoustic detection model using an ensemble of Convolutional Neural Network models in order to automate the non-acoustic detection. The proposed model is trained to classify targets as 4 classes which are submarines, flag buoys, lighted buoys, small boats. Based on the numerical study with 10,287 images, we confirm the proposed model can achieve 91.5 % test accuracy for the non-acoustic detection of submarines.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2002년도 추계학술대회논문집
• /
• pp.711-714
• /
• 2002

At the lift-off condition, the combustion and Jet noise of launch vehicle produces a severe acoustic environment and the acoustic loads may be damaging to paylaod and equipments. Prediction of the acoustic environment is thus needed to support the load-resistive design and test-qualification of components. Currently, such a high frequency problem is usually dealt with by using the SEA technique, for which the assumptions should match reasonably well with the vibro-acoustic condition of system. The subsystems of SEA model was composed of 16 flat plates, 8 L-shaped beams, and 2 acoustic cavities. The frequency range was 400 Hz - 4 kHz considering the modal parameter. The experiment was performed in a high intensity acoustic chamber, in which the diffuse acoustic field was assured. By comparing the SEA analysis and the experiments, the error less than 5 dB was observed.

• 대한기계학회:학술대회논문집
• /
• 대한기계학회 2003년도 춘계학술대회
• /
• pp.1827-1832
• /
• 2003

In this research, the simulation method for acoustic sounds by a uniform flow around a two-dimensional circular cylinder by using the finite difference lattice Boltzmann model is explained. To begin with, we examine the boundary condition which determined with the distribution function $f_i^{(0)}$ concerning with density, velocity and internal energy at boundary node. Very small acoustic pressure fluctuation, with same frequency as that of Karman vortex street, is compared with the pressure fluctuation around a circular cylinder. The acoustic sound' propagation velocity shows that acoustic approa ching the upstream, due to the Doppler effect in the uniform flow, slowly propagated. For the do wnstream, on the other hand, it quickly propagates. It is also apparently the size of sound pressure was proportional to the central distance $r^{-1/2}$ of the circular cylinder. The lattice BGK model for compressible fluids is shown to be one of powerful tool for simulation of gas flows.

• 말소리와 음성과학
• /
• 제9권2호
• /
• pp.95-101
• /
• 2017

This paper proposes a new method to train Deep Neural Network (DNN)-based acoustic models for speech recognition of native and foreign speakers. The proposed method consists of determining multi-set state clusters with various acoustic properties, training a DNN-based acoustic model, and recognizing speech based on the model. In the proposed method, hidden nodes of DNN are shared, but output nodes are separated to accommodate different acoustic properties for native and foreign speech. In an English speech recognition task for speakers of Korean and English respectively, the proposed method is shown to slightly improve recognition accuracy compared to the conventional multi-condition training method.

• 대한음성학회지:말소리
• /
• 제68권
• /
• pp.33-47
• /
• 2008

In this paper, we propose a hybrid acoustic and pronunciation model adaptation method based on context dependency for Korean-English speech recognition. The proposed method is performed as follows. First, in order to derive pronunciation variant rules, an n-best phoneme sequence is obtained by phone recognition. Second, we decompose each rule into a context independent (CI) or a context dependent (CD) one. To this end, it is assumed that a different phoneme structure between Korean and English makes CI pronunciation variabilities while coarticulation effects are related to CD pronunciation variabilities. Finally, we perform an acoustic model adaptation and a pronunciation model adaptation for CI and CD pronunciation variabilities, respectively. It is shown from the Korean-English speech recognition experiments that the average word error rate (WER) is decreased by 36.0% when compared to the baseline that does not include any adaptation. In addition, the proposed method has a lower average WER than either the acoustic model adaptation or the pronunciation model adaptation.

• Structural Engineering and Mechanics
• /
• 제64권1호
• /
• pp.93-107
• /
• 2017

The article presents the vibration and acoustic responses of un-baffled doubly curved laminated composite panel structure under the excitation of a harmonic point load. The structural responses are obtained using a simulation model via ANSYS including the effect various geometries (cylindrical, elliptical, spherical and hyperboloid). Initially, the model has been established by solving adequate number of available examples to show the convergence and comparison behaviour of the natural frequencies. Further, the acoustic responses are obtained using an indirect boundary element approach for the coupled fluid-structure analysis in LMS Virtual.lab by importing the natural frequency values. Subsequently, the values for the sound power level are computed using the present numerical model and compared with that of the available published results and in-house experimentally obtained data. Further, the acoustic responses (mean-square velocity, radiation efficiency and sound power level) of the doubly curved layered structures are evaluated using the current simulation model via several numerical experimentations for different structural parameters and corresponding discussions are provided in detail.

• 한국소음진동공학회논문집
• /
• 제24권12호
• /
• pp.969-976
• /
• 2014

Nowadays, it is required for naval ships to estimate 3D underwater radiated noise pattern in all direction at peak frequencies of hull vibration for the reduction of being detected and doing the effective operation. For this purpose, the numerical method has to be developed to calculate 3D underwater radiated noise pattern with experimental data. It is very difficult to obtain the experimental data for the real ship. Alternative to get the experimental results is to use NAH(near-field acoustic holography) in acoustic tank with experimental model. Application of NAH in acoustic tank for the experimental model needs some investigation of reflection wave from the wall of the acoustic tank and unmeasured zone of the experimental model due to the supporting structure for it. In this study, the effect of reflection wave in the acoustic tank and unmeasured area of the experimental model when using the NAH was investigated with experiment and numerical model. From these, it is known for the error due to reflection wave can be reduced when the distance between the measurement plane and source is being shorten. Also, unmeasured area of the experimental model gives rise to some error in the estimation of the far-field acoustic pressure.

• 한국추진공학회지
• /
• 제23권2호
• /
• pp.38-45
• /
• 2019

본 연구에서는 항공용 가스터빈의 연소실에서의 연소불안정 해석을 위한 고유값 도출을 목적으로 하는 1D 네트워크 모델을 개발하였다. 모델은 면적 변화가 있는 음향 네트워크 요소들 사이의 각종 지배 방정식을 통하여 개발되었고, 이를 이용하여 현재 개발 중인 복잡한 유로 형상을 갖는 실제 항공용 가스터빈 연소기에서의 음향장 해석에 적용되었다. 본 모델을 통하여 도출된 음향장 해석 결과는 3차원 유한요소해석 기반의 헬름홀츠 솔버의 계산 결과와 비교하였다.

• 한국추진공학회:학술대회논문집
• /
• 한국추진공학회 2009년도 제33회 추계학술대회논문집
• /
• pp.399-402
• /
• 2009

로켓엔진 연소기의 음향 불안정 억제를 위해 헬름홀츠 공명기의 음향학적 최적 설계 설계조건을 실험적, 해석적으로 연구하였다. 표준 음향시험 절차에 따라 음향파관에 공명기를 장착하여 음향파응답 신호를 측정하였다. 측정된 신호를 토대로, 흡음계수를 계산하였고, 이로부터 공명기의 음향감쇠 성능을 파악하였다. 헬름홀츠 공명기를 스프링-댐퍼 시스템으로 상사하여, 음향학적 감쇠성능을 이해하였다. 헬름홀츠 공명기의 설계변수로 목의 길이와 공동의 크기 등을 선택하였다. 공동의 크기가 커질수록 흡음성능이 증가하였고, 목 길이가 길어질수록 흡음성능이 감소하였다.

• 한국음향학회지
• /
• 제18권8호
• /
• pp.3-8
• /
• 1999

문맥독립형 음향모델을 채택하고 있는 기존의 가변어휘 음성인식기는 주변환경에 따른 음소의 변화를 모델링 할 수 없었다. 이러한 문제를 해결하기 위해서는 변이음을 이용한 문맥의존형 음향모델을 사용해야 한다. 본 논문은 가변어휘 음성인식기의 음향모델을 효과적으로 개선하기 위하여 적용한 방법에 대해서 기술하고 있다. 즉, 음향모델의 개선은 엔트로피를 이용한 군집화 기법을 적용하여 변이음의 개수를 변경시키면서 최적의 변이음 모델을 추출하는 방법을 사용하였다. 개선된 모델에 대한 성능은 POW(Phonetically Optimized Words) 3848 DB 및 SNR이 크게 다른 2종류의 PC168 DB를 이용하여 훈련 및 인식 실험을 수행하면서 평가하였다. 결론적으로 변이음의 개수를 낮추면서도 인식 성능의 저하를 가져오지 않는 최적의 변이음 모델을 얻을 수 있었으며 PC168 DB를 이용한 인식실험을 통하여 확인할 수 있었다.