• 한국전산구조공학회논문집
• /
• 제16권2호
• /
• pp.115-124
• /
• 2003

이 논문에서는 기존의 보-거더 구조계 주차장 구조물에 대한 차량하중영향 연구를 토대로, 플랫 슬래브 구조계에서 차량하중영향에 대한 연구를 수행하였다. 먼저, 최대부재력을 일으키는 차량하중의 적용을 위해 플랫 슬래브의 주요 설계지점에 대한 영향면을 구성하였으며, 플랫 슬래브의 등가차량하중계수를 인공 신경망기법을 이용하여, 슬래브 두께, 지판 두께, 지판 크기, 슬래브의 단변, 장변 길이 등 주요구조변수로 제시하였다. 사용된 신경망의 훈련은 많은 패턴수를 갖는 비선형 회귀분석에 적합한 Levenberg-Marquardt 알고리즘을 이용하였으며 해석결과와 인공 신경망의 출력의 비교를 통해 알고리즘의 유효성을 검증하였다. 플랫 슬래브 구조계의 등가차량하중계수를 살펴보면, 보-거더 구조계의 경우와 유사하게 주열대와 중간대의 정모멘트 부재력에서 차량하중에 매우 취약함을 알 수 있었다.

• Smart Structures and Systems
• /
• 제25권5호
• /
• pp.605-617
• /
• 2020

The existence of damages in structures causes changes in the physical properties by reducing the modal parameters. In this paper, we develop a two-stages approach based on normalized Modal Strain Energy Damage Indicator (nMSEDI) for quick applications to predict the location of damage. A two-dimensional IsoGeometric Analysis (2D-IGA), Machine Learning Algorithm (MLA) and optimization techniques are combined to create a new tool. In the first stage, we introduce a modified damage identification technique based on frequencies using nMSEDI to locate the potential of damaged elements. In the second stage, after eliminating the healthy elements, the damage index values from nMSEDI are considered as input in the damage quantification algorithm. The hybrid of Teaching-Learning-Based Optimization (TLBO) with Artificial Neural Network (ANN) and Particle Swarm Optimization (PSO) are used along with nMSEDI. The objective of TLBO is to estimate the parameters of PSO-ANN to find a good training based on actual damage and estimated damage. The IGA model is updated using experimental results based on stiffness and mass matrix using the difference between calculated and measured frequencies as objective function. The feasibility and efficiency of nMSEDI-PSO-ANN after finding the best parameters by TLBO are demonstrated through the comparison with nMSEDI-IGA for different scenarios. The result of the analyses indicates that the proposed approach can be used to determine correctly the severity of damage in beam structures.

• Journal of the Optical Society of Korea
• /
• 제17권4호
• /
• pp.289-295
• /
• 2013

Although low-level laser therapy (LLLT) has been a valuable therapeutic technology in the clinic, its efficacy may be reduced in deep tissue layers due to strong light scattering which limits the photon density. In order to enhance the photon density in deep tissue layers, this study developed an optical tissue clearing (OTC) laser probe (OTCLP) system which can utilize four different OTC methods: 1) tissue temperature control from 40 to $10^{\circ}C$; 2) laser pulse frequency from 5 to 30 Hz; 3) glycerol injection at a local region; and 4) a combination of the aforementioned three methods. The efficacy of the OTC methods was evaluated and compared by investigating laser beam profiles in ex-vivo porcine skin samples. Results demonstrated that total (peak) intensity at full width at half maximum of laser beam profile when compared to control data was increased: 1) 1.21(1.39)-fold at $10^{\circ}C$; 2) 1.22 (1.49)-fold at a laser pulse frequency of 5 Hz; 3) 1.64 (2.41)-fold with 95% glycerol injection; 4) 1.86 (3.4)-fold with the combination method. In conclusion, the OTCLP system successfully improved the laser photon density in deep tissue layers and may be utilized as a useful tool in LLLT by increasing laser photon density.

• 한국분무공학회지
• /
• 제28권1호
• /
• pp.1-9
• /
• 2023

In the cylinder of gasoline direct injection engines, the spray targeting from injectors is of great significance for fuel consumption and pollutant emissions. The automotive industry is putting a lot of effort into improving injector targeting accuracy. To improve the targeting accuracy of injectors, it is necessary to develop models that can predict the spray targeting positions. When developing spray targeting models, the most used technique is computational fluid dynamics (CFD). Recently, due to the superiority of machine learning in prediction accuracy, the application of machine learning in this field is also receiving constant attention. The purpose of this study is to build a machine learning model that can accurately predict spray targeting based on the design parameters of injectors. To achieve this goal, this study firstly used laser sheet beam visualization equipment to obtain many spray cross-sectional images of injectors with different parameters at different injection pressures and measurement planes. The spray images were processed by MATLAB code to get the targeting coordinates of sprays. A total of four models were used for the prediction of spray targeting coordinates, namely ANN, LSTM, Conv1D and Conv1D & LSTM. Features fed into the machine learning model include injector design parameters, injection conditions, and measurement planes. Labels to be output from the model are spray targeting coordinates. In addition, the spray data of 7 injectors were used for model training, and the spray data of the remaining one injector were used for model performance verification. Finally, the prediction performance of the model was evaluated by R² and RMSE. It is found that the Conv1D&LSTM model has the highest accuracy in predicting the spray targeting coordinates, which can reach 98%. In addition, the prediction bias of the model becomes larger as the distance from the injector tip increases.

• 한국주거학회논문집
• /
• 제26권1호
• /
• pp.1-10
• /
• 2015

In line with the recent public concern on the environmental issues in building industry, there has been a rise in demand for a healthy, sustainable housing environment in South Korea. In order to achieve a healthy environment in residential buildings, considerable efforts have been made in a wide range of sectors. Among others, the development of the certification schemes to promote environment-friendly planning and building construction is remarkable. In urban South Korea, recently built houses tend to be significantly high-rise, high-density buildings. Global warming has brought about drastic climate change and continued to increase the average annual temperature year by year. These changes should be well reflected on the government's implementation of the building environmental assessment system. For guidance, therefore, this study looks to the case of Hong Kong which is well known for high-density housing development and subtropical climate conditions. It compares the features of the green building certification schemes for newly developed multifamily housing in two regions, namely HK-BEAM in Hong Kong and G-SEED in South Korea. Based on the findings, it argues that the G-SEED implementor should have expanded roles in providing training programs and follow-up services in collaboration with the certification authorities. It is also argued that G-SEED professionals should be involved in the early stages of design processes, and training programs and licence systems to produce green building professionals should be developed. Finally, it points out that the assessment indicators should be more detailed and diversified.

• Journal of Communications and Networks
• /
• 제18권6호
• /
• pp.892-901
• /
• 2016

In the multi-user multiple-input multiple-output (MIMO) system, orthogonal random beamforming (ORBF) scheme is proposed to serve multiple users simultaneously in order to achieve the multi-user diversity gain. The opportunistic space-division multiple access system (OSDMA-S) scheme performs multiple weighting matrices during the training phase and chooses the best weighting matrix to be used to broadcast data during the transmitting phase. The OSDMA-S scheme works better than the original ORBF by decreasing the inter-user interference during the transmitting phase. To save more time in the training phase, a partly random multiple weighting matrices selection scheme is proposed in this paper. In our proposed scheme, the Base Station does not need to use several unitary matrices to broadcast pilot symbol. Actually, only one broadcasting operation is needed. Each subscriber generates several virtual equivalent channels with a set of pre-saved unitary matrices and the channel status information gained from the broadcasting operation. The signal-to-interference and noise ratio (SINR) of each beam in each virtual equivalent channel is calculated and fed back to the base station for the weighting matrix selection and multi-user scheduling. According to the theoretical analysis, the proposed scheme relatively expands the transmitting phase and reduces the interactive complexity between the Base Station and subscribers. The asymptotic analysis and the simulation results show that the proposed scheme improves the throughput performance of the multi-user MIMO system.

• 한국운동역학회지
• /
• 제18권1호
• /
• pp.137-146
• /
• 2008

본 연구는 현 국가대표 선수들을 대상으로 평균대에서 백핸드 수완 동작의 성공과 실패를 운동학적 분석을 통해 비교하여 기술의 실수 요인을 규명하여 지도자 및 선수들에게 과학적으로 유용한 정보를 제공하고 경기력 향상에 기여 하는데 연구의 목적을 두었다. 신체중심 변위와 속도 변화에서 실수 동작에 영향을 미치는 중요한 요인은 좌우 신체중심 속도 변화와 수평과 수직 속도 변화로 나타났고 좌우 가속도 변화는 성공시 보다 실패시 동작이 더 크게 증가하였으며 E3과 E5에서 수평과 수직 가속도가 뒤 공중돌기와 착지구간에 영향을 미치는 중요한 구간으로 나타났다. 성공과 실패시 차이가 나타난 각속도 변화는 머리와 견관절 결과에서 두드러지게 나타났으며 머리와 견관절 각가속도를 가장 크게 해야 하는 순간은 E4라고 판단되며 이 구간에서 신체가 굴곡된 자세에서 순간적으로 신전하는 동작을 취할 때 머리와 견관절, 고관절 등 각 관절의 각가속도를 크게 증가시켜 공중 동작의 체공시간과 회전 반경을 더욱 원활하게 할 수 있다.

• 공학교육연구
• /
• 제10권4호
• /
• pp.17-28
• /
• 2007

본 논문에서는 지역균형발전의 일환으로 실시된 지방대학 혁신역량 강화사업(New University for Regional Innovation, NURI) 중 부산권 기계부품산업 혁신인력 양성사업단(Busan Educational Alliance of Mechanical Engineering, BEAM)에서 시행중인 지역대학간 공동강의 시스템인 첨단기계부품 특화교육트랙을 통한 교육효과와 교과과정 개선의 질적 향상방안을 제시하였다. 트랙은 사업단 소속 4개 대학의 기계공학부(과)가 공동주관하며, 각 대학의 특화된 교육과정(첨단기계, 환경기계, 해양기계, 기반기계)을 중심으로 상호학점 인정을 통한 교과과정으로 편성되었다. 2005년 동계학기부터 현재까지 3차례의 계절학기를 통하여 실시되었으며, 2년간 총 486명의 학생이 각 트랙과목을 수강하였다. 그 결과 새로운 시대적 요구사항과 학습의 효율성을 동시에 수용하는 성과중심 및 수요자중심의 교육내용으로 국내 최초의 전공별 지역공동강의 제도를 정착시켰으며, 트랙 실시 이전인 2005년 대비 8.5%의 졸업생 취업률 향상과 기업체 임직원을 대상으로 실시된 설문조사에서 전년대비 9.4%의 수요자 만족도 향상의 성과를 거두었다. 또한, 지역대학간의 교류를 통한 인적 네트워크의 확대라는 긍정적인 결과를 얻을 수 있었다.

• 한국음향학회지
• /
• 제40권5호
• /
• pp.530-536
• /
• 2021

최근 들어 딥러닝의 발달로 인해 Hidden Markov Model(HMM)을 사용하지 않고 음성 신화와 단어를 직접 매핑하여 학습하는 end-to-end 음성인식 방법이 각광을 받고 있으며 그 중에서도 conformer가 가장 좋은 성능을 보이고 있다. 하지만 end-to-end 음성인식 방법은 현재 시점에서 어떤 자소 또는 단어가 나타날지에 대한 확률에 대해서만 초점을 두고 있다. 그 이후의 디코딩 과정은 현재 시점에서 가장 높은 확률을 가지는 자소를 출력하거나 빔 탐색을 사용하며 이러한 방식은 모델이 출력하는 확률 분포에 따라 최종 결과에 큰 영향을 받게 된다. 또한 end-to-end 음성인식방식은 전통적인 음성인식 방법과 비교 했을 때 구조적인 문제로 인해 외부 발음열 정보와 언어 모델의 정보를 사용하지 못한다. 따라서 학습 자료에 없는 발음열 변환 규칙에 대한 대응이 쉽지 않다. 따라서 본 논문에서는 발음열 정보를 담고 있는 Lexicon transducer(L transducer)를 이용한 conformer의 디코딩 방법을 제안한다. 한국어 데이터 셋 270 h에 대해 자소 기반 conformer의 빔 탐색 결과와 음소 기반 conformer에 L transducer를 적용한 결과를 비교 평가하였다. 학습자료에 등장하지 않는 단어가 포함된 테스트 셋에 대해 자소 기반 conformer는 3.8 %의 음절 오류율을 보였으며 음소 기반 conformer는 3.4 %의 음절 오류율을 보였다.

• Smart Structures and Systems
• /
• 제4권5호
• /
• pp.583-603
• /
• 2008

In this study, a real-time damage detection method using output-only acceleration signals and artificial neural networks (ANN) is developed to monitor the occurrence of damage and the location of damage in structures. A theoretical approach of an ANN algorithm that uses acceleration signals to detect changes in structural parameters in real-time is newly designed. Cross-covariance functions of two acceleration responses measured before and after damage at two different sensor locations are selected as the features representing the structural conditions. By means of the acceleration features, multiple neural networks are trained for a series of potential loading patterns and damage scenarios of the target structure for which its actual loading history and structural conditions are unknown. The feasibility of the proposed method is evaluated using a numerical beam model under the effect of model uncertainty due to the variability of impulse excitation patterns used for training neural networks. The practicality of the method is also evaluated from laboratory-model tests on free-free beams for which acceleration responses were measured for several damage cases.