• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.26 no.4
• /
• pp.213-221
• /
• 2013

This paper introduces a mesh continuation scheme for a one-dimensional inverse medium problem to reconstruct the spatial distribution of elastic wave velocities in heterogeneous semi-infinite solid domains. To formulate the inverse problem, perfectly-matched-layers(PMLs) are introduced as wave-absorbing boundaries that surround the finite computational domain truncated from the originally semi-infinite extent. To tackle the inverse problem in the PML-truncated domain, a partial-differential-equations(PDE)-constrained optimization approach is utilized, where a least-squares misfit between calculated and measured surface responses is minimized under the constraint of PML-endowed wave equations. The optimization problem iteratively solves for the unknown wave velocities with their updates calculated by Fletcher-Reeves conjugate gradient algorithms. The optimization is performed using a mesh continuation scheme through which the wave velocity profile is reconstructed in successively denser mesh conditions. Numerical results showed the robust performance of the mesh continuation scheme in reconstructing target wave velocity profile in a layered heterogeneous solid domain.

• Journal of Advanced Marine Engineering and Technology
• /
• v.39 no.1
• /
• pp.8-12
• /
• 2015

Non-esterified bio-diesel fuel saves cost by no esterified process and its performance was more similar to diesel oil than esterified bio-diesel fuel when the fuel blended 95% diesel oil and 5% it was used on diesel engine with electronic control system. A performance optimization is necessary for application of non-esterified bio-diesel fuel blended with diesel oil 95% on the latest diesel engine. In this study, test using fractional factorial design was accomplished at 25% and 50% partial load in order to evaluate influence of controllable 6 factors on responses such as specific fuel consumption, nitrogen oxides and coefficiency of variation of indicated mean effective pressure as basic experiment for performance optimization of this fuel. It is cleared that the injection timing and common rail pressure of 6 factors are mainly effective and its effect level is different according to load.

• Journal of Korea Society of Industrial Information Systems
• /
• v.29 no.2
• /
• pp.153-167
• /
• 2024

DMAIC is the most familiar problem-solving methodology to battery manufacturing-related engineers, but continuous problems such as task delay, insufficient performance, and partial optimization are occurring due to indiscriminate application to various tasks of battery production system developers. In order to secure an "optimized model for DMAIC methodology" that can effectively respond to battery production system developers' tasks, a three-stage research model was used to derive the required characteristics of the production system developer task methodology, analyze the suitability of DMAIC, and conduct optimization modeling by supplementing the shortcomings. It was confirmed that the DMAIC methodology can be more suitable by applying the "system structural seven-step methodology", which is the result of this study, to developer tasks. It is expected that it will be applied to various industrial fields in the future by making it easier to learn and allowing differentiated operations according to the characteristics of various industries.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.17 no.4
• /
• pp.213-220
• /
• 2024

This paper proposes an optimization strategy for performing Federated Learning between devices and edge clouds using WebAssembly (WASM). The proposed strategy aims to maximize efficiency by conducting partial training on devices and the remaining training on edge clouds. Specifically, it mathematically describes and evaluates methods to optimize data transfer between GPU memory segments and the overlapping of computational tasks to reduce overall training time and improve GPU utilization. Through various experimental scenarios, we confirmed that asynchronous data transfer and task overlap significantly reduce training time, enhance GPU utilization, and improve model accuracy. In scenarios where all optimization techniques were applied, training time was reduced by 47%, GPU utilization improved to 91.2%, and model accuracy increased to 89.5%. These results demonstrate that asynchronous data transfer and task overlap effectively reduce GPU idle time and alleviate bottlenecks. This study is expected to contribute to the performance optimization of Federated Learning systems in the future.

• The Journal of Korean Association of Computer Education
• /
• v.14 no.2
• /
• pp.83-94
• /
• 2011

In the biomedical domain, the longest matching method is frequently used for recognizing named entity written in the literature. This method uses a dictionary as a resource for named entity recognition. If there exist appropriated dictionary about target domain, the longest matching method has the advantage of being able to recognize the entities of target domain quickly and exactly. However, the longest matching method is difficult to recognize the enumerated named entities, because these entities are frequently expressed as being omitted some words. In order to resolve this problem, we propose the partial matching method using a dictionary. The proposed method makes several candidate entities on the assumption that the ellipses may be included. After that, the method selects the most valid one among candidate entities through the optimization algorithm. We tested the longest and partial matching method about HLA entities: HLA gene, antigen, and allele entities, which are frequently enumerated among biomedical entities. As preparing for named entity recognition, we built two new resource, extended dictionary and tag-based dictionary about HLA entities. And later, we performed the longest and partial matching method using each dictionary. According to our experiment result, the longest matching method was effective in recognizing HLA antigen entities, in which the ellipses are rare, and the partial matching method was effective in recognizing HLA gene and allele entities, in which the ellipses are frequent. Especially, the partial matching method had a high F-score 95.59% about HLA alleles.

• Journal of Distribution Science
• /
• v.15 no.8
• /
• pp.15-28
• /
• 2017

Purpose - Many researchers analyze VMI as a supply chain collaboration program to reveal its true value. Most of them focus on the dyadic relationship in two stage supply chain systems. This study examines the effect of VMI when it is applied to the different parts of three stage supply chain systems. Research design, data, and methodology - Based on three stage supply chain, this study compares three different systems including full VMI, partial VMI, and non-VMI by using mathematical models. The performances of three systems are compared with the numerical examples of the proposed supply chain models. Results - The numerical examples reveal that full VMI where the manufacturer controls inventories at all stages outperforms any other systems in terms of the system profit and enables all individual members to gain greater profits than non-VMI. Meanwhile, under partial VMI where VMI is implemented between the wholesaler and retailer, only these two members improve their performances and the manufacturer who does not belong to VMI makes less profit than even under non-VMI. This study also examines the impact of market size and profit margin on the system performance. Conclusions - The result of this study supports the common belief that VMI secures the best result when it is applied to the entire supply chain system. The additional findings from the numerical analysis are discussed.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.16 no.4
• /
• pp.281-292
• /
• 2016

The proposed scheme is based on the modified perturb and observe (P&O) algorithm combined with the sliding mode technique. A modified P&O algorithm based sliding mode controller is developed to study the effects of partial shade, temperature, and insolation on the performance of maximum power point tracking (MPPT) used in photovoltaic (PV) systems. Under partially shaded conditions and temperature, the energy conversion efficiency of a PV array is very low, leading to significant power losses. Consequently, increasing efficiency by means of MPPT is particularly important. Conventional techniques are easy to implement but produce oscillations at MPP. The proposed method is applied to a model to simulate the performance of the PV system for solar energy usage, which is compared to the conventional methods under non-uniform insolation improving the PV system utilization efficiency and allowing optimization of the system performance. The modified perturb and observe sliding mode controller successfully overcomes the issues presented by non-uniform conditions and tracks the global MPP. Compared to MPPT techniques, the proposed technique is more efficient; it produces less oscillation at MPP in the steady state, and provides more precise tracking.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.31 no.6
• /
• pp.351-355
• /
• 2019

Load resistance factors for the limit state design of vertical caisson breakwaters are presented. Reliability analysis of 16 breakwaters in nationwide ports was conducted to calculate the partial safety factors and they were converted into load and resistance factors. The final load resistance factor was calibrated by applying the optimization technique to the individually calculated load resistance factors. Finally, the breakwater was redesigned using the optimal load resistance factor and verified whether the target level was met. The load resistance factor according to the change of the target reliability level is presented to facilitate the limit state design of breakwater.

• Proceedings of the KIEE Conference
• /
• 2006.10a
• /
• pp.145-146
• /
• 2006

본 논문은 부분방전(PD: Partial Discharge)의 패턴인식 확률 극대화를 목적으로 신경망(NN: Neural Network) 파라미터 중에서 은닉층 뉴런의 수, 모멘텀(momentum)의 Step size와 Decay rate 를 최적화하기 위하여 유전 알고리즘(GA: Genetic Algonthm)을 적응하였다. 실험적 연구의 대상으로서, GIS(Gas Insulated Switchgear)사고의 주요 원인으로 보고되어있는 결함들을 인위적으로 모의한 16개 Test cell을 이용하여 부분방전을 발생시켰다. 부분방전 신호는 본 연구팀이 개발한 센서를 이용하여 검출되어 데이터베이스가 구축되어 그로부터 추출된 학습 데이터들의 학습에 다음과 같은 5가지 신경망 모델이 적응되었다: Multilayer Perception (MLP), Jordan-Elman Network (JEN), Recurrent Network (RN), Self-Organizing Feature Map (SOFM), Time-Lag Recurrent Network (TLRN). 유전 알고리즘 적용 효율성을 분석하기 위하여 동일한 데이터를 이용하여 다음과 같은 두 가지 방법을 적용한 결과를 상호 비교하였다. 우선 상기 선택된 모델만 적용하였고 다근 하나는 상기 모델과 Genetic Algorithm이 동시에 적용되었다. 모든 모델에 대하여 학습오차와 패턴 분류 확률을 비교한 결과, 유전 알고리즘 적응 시 부분방전 패턴인식 확률이 향상되었음이 확인되어 향후 신뢰성 있는 GIS 부분방전 진단기술에 활용될 수 있을 것으로 사료된다.

• Korean Journal of Materials Research
• /
• v.16 no.11
• /
• pp.668-675
• /
• 2006

In the partial squeeze casting process, the filling behavior of liquid metal and solidification pattern in thick area have significant influence on the quality of casting products and die life. For the optimal casting design of automobile transmission gear housing, various analyses were performed in this study by using computer simulation code, MAGMAsoft and the simulation results were compared and analyzed with experimental results. By air pressure criteria, internal porosities caused by air entrap during the mold filling were predicted and reduced remarkably by modification of gating system. Also, optimal squeeze-time lag to apply partial squeeze pin in thick area was calculated and the castings was free from shrinkage defects with the result of solidification analysis. Consequently, casting design for automobile transmission gear housing was optimized and approved by Computer Tomography.