• Proceedings of the Korea Information Processing Society Conference
• /
• 2004.05a
• /
• pp.1651-1654
• /
• 2004

파이프라인과 슈퍼스칼라 방식이 일반화된 시스템 구조 하에서, 분기 명령어는 시스템 전체적인 성능에 중요한 영향을 미친다. 특히 분기 예측이 실패했을 경우, 잘못된 분기 예측으로 인한 페널티가 발생한다는 점에서 분기 예측의 정확도에 대한 중요성은 크다고 할 수 있다. 본 논문에서는 분기 예측의 정확도를 높이기 위해서, 분기 예측과 관련된 신경망을 구축하여 이를 통해 분기 예측에 필요한 각 요소별 가중치의 변화를 분석하고, 이를 분기 예측에 새롭게 반영하고자 한다. 본 논문에서는 이를 위해 실행 구동 방식의 시뮬레이터인 SimpleScalar를 통하여 모의 실험을 수행하였으며, 실험 결과 본 논문에서 제시한 새로운 기법이 기존의 일반적인 이단계 적응형 분기 예측 기법이나 gshare 기법에 비하여 더 우수한 결과를 보였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2009.11a
• /
• pp.55-56
• /
• 2009

내장형 프로세서에 대한 기능요구사항이 날로 증가함에 따라 데이터 처리량을 늘리기 위한 많은 연구들이 지속되어 왔으며, 그중 파이프라인의 컨트롤 해저드로 인한 성능저하를 최소화하기 위한 분기 예측 기법이 다양한 방식으로 제안되어 왔다. 본 논문에서는 분기예측 방법으로서 구현이 간단하고 분기 예측률이 높은 BTB 방식을 32비트 프로세서에 적용하고, 해당 프로세서를 사용하는 SoC 플랫폼을 구성하여 분기예측기법 사용으로 인한 성능향상을 측정하고, 0.18um ASIC 공정을 적용하여 SoC 플랫폼을 구현한 결과를 제시한다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.07a
• /
• pp.335-338
• /
• 2023

국방 무기체계의 운용유지를 위해서는 숙련자에 의한 신뢰성있는 정비 지원이 필요하다. 특히, 고도의 기술력을 바탕으로 연구/제작된 해군 무기체계를 유지하기 위해서는 이와같은 정비 지원이 무엇보다 중요하다. 해군에서는 효과적인 정비지원을 위해 수개의 정비지원부대를 조직하여 운용하고 있다. 원활한 정비지원부대의 운용을 위해 다년간 기술력을 축적한 정비인원의 중도 이탈을 예방하는 것이 요구되므로, 본 논문에서는 머신러닝을 적용하여 해군 정비지원부대의 퇴직자 예측 모델을 제안하였다. 정비인력의 만족도와 관계가 있을 것으로 예상되는 봉급, 특근율 등을 변수로 사용하였고, F1 Score를 통해 모델의 성능을 평가한 결과 0.7이상의 높은 성능을 보였다. 이 모델을 통해 조기 퇴직이 예상되는 그룹의 공통 개선소요를 파악하여 사전 조치가 가능할 것으로 판단하였다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.07a
• /
• pp.339-342
• /
• 2023

해군에서 운용하는 각 함정은 여러 가지 무기체계를 동시에 탑재하고 있는, 고가의 복합무기체계이다. 주어진 기간동안 효과적인 정비를 수행함으로써, 가동률을 극대화하는 것이말로 복합무기체계인 함정을 효과적으로 운용하는 방법이며, 경제적인 국방운용이라 부를 수 있을 것이다. 정비인시는 여러 무기체계가 동시에 탑재된 복합무기체계 정비의 핵심이다. 정비인시를 정확하게 알고 있어야만, 제한된 정비기간을 각 무기체계에 정확히 할당할 수 있을 것이며, 최적의 요소에 대한 정비가 수행될 수 있을 것이다. 본 연구에서는 해군에서 운용중인 특정장비에 대한 약 10년간의 정비자료를 이용하여 정비인시를 예측하는 모델을 제안하였다. 모델의 성능은 R2 Score를 통해 0.69의 준수한 수치를 보였다. 이 모델을 통해 조금더 세밀하고, 정확한 정비인시 예측과 정비계획 수립이 가능하리라 판단하였다.

• Structural Engineering and Mechanics
• /
• v.66 no.4
• /
• pp.505-513
• /
• 2018

Steam flooding is widely used in heavy oil reservoir with coupling effects among the formation temperature change, fluid flow and solid deformation. The effective stress, porosity and permeability in this process can be affected by the multi-physical coupling of thermal, hydraulic and mechanical processes (THM), resulting in a complex interaction of geomechanical effects and multiphase flow in the porous media. Quantification of the state of deformation and stress in the reservoir is therefore essential for the correct prediction of reservoir efficiency and productivity. This paper presents a coupled fluid flow, thermal and geomechanical model employing a program (MATLAB interface code), which was developed to couple conventional reservoir (ECLIPSE) and geomechanical (ABAQUS) simulators for coupled THM processes in multiphase reservoir modeling. In each simulation cycle, time dependent reservoir pressure and temperature fields obtained from three dimensional compositional reservoir models were transferred into finite element reservoir geomechanical models in ABAQUS as multi-phase flow in deforming reservoirs cannot be performed within ABAQUS and new porosity and permeability are obtained using volumetric strains for the next analysis step. Finally, the proposed approach is illustrated on a complex coupled problem related to steam flooding in an oil reservoir. The reservoir coupled study showed that permeability and porosity increase during the injection scenario and increasing rate around injection wells exceed those of other similar comparable cases. Also, during injection, the uplift occurred very fast just above the injection wells resulting in plastic deformation.

• KIISE Transactions on Computing Practices
• /
• v.22 no.12
• /
• pp.681-686
• /
• 2016

Software defect prediction is an important factor in efficient project management and success. The severity of the defect usually determines the degree to which the project is affected. However, existing studies focus only on the presence or absence of a defect and not the severity of defect. In this study, we proposed an ensemble model using FCM based on defect severity. The severity of the defect of NASA data set's PC4 was reclassified. To select the input column that affected the severity of the defect, we extracted the important defect factor of the data set using Random Forest (RF). We evaluated the performance of the model by changing the parameters in the 10-fold cross-validation. The evaluation results were as follows. First, defect severities were reclassified from 58, 40, 80 to 30, 20, 128. Second, BRANCH_COUNT was an important input column for the degree of severity in terms of accuracy and node impurities. Third, smaller tree number led to more variables for good performance.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.44 no.2
• /
• pp.149-153
• /
• 1999

Prediction of rice developmental stage is necessary for proper crop management and a prerequisite for growth simulation as well. The objectives of the present study were to find out the relationship between the plastochrone index(PI) and the developmental index(DVI) estimated by non-parametric phenology model which simulates the duration from seedling emergence(DVI=0) to heading(DVI=l) by employing daily mean air temperature and daylength as predictor variables, and to confirm the correspondency of developmental indice to panicle developmental stages based on this relationship. Four japonica rice cultivars, Kwanakbyeo, Sangpungbyeo, Dongjinbyeo, and Palgumbyeo which range from very early to very late in maturity, were grown by sowing directly in dry paddy field five times at an interval of two weeks. Data for seedling emergence, leaf appearance, differentiation stage of primary rachis branch and heading were collected. The non-parametric phenology model predicted well the duration from seedling emergence to heading with errors of less than three days in all sowings and cultivars. PI was calculated for every leaf appearance and related to the developmental index estimated for corresponding PI. The stepwise polynomial analysis produced highly significant square-rooted cubic or biquadratic equations depending on cultivars, and highly significant square-rooted biquadratic equation for pooled data across cultivars without any considerable reduction in accuracy compared to that for each cultivar. To confirm the applicability of this equation in predicting the panicle developmental stage, DVI at differentiation stage of primary rachis branch primordium was calculated by substituting PI with 82 corresponding to this stage, and the duration reaching this DVI from seedling emergence was estimated. The estimated duration revealed a good agreement with that observed in all sowings and cultivars. The deviations between the estimated and the observed were not greater than three days, and significant difference in accuracy was not found for predicting this developmental stage between those equations derived for each cultivar and for pooled data across all cultivars tested.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.39 no.5
• /
• pp.391-399
• /
• 2011

The purpose of this paper is to predict the shock absorbing characteristics of the aluminum honeycomb in a lunar lander. Aluminum honeycomb has been used for shock absorbers of lunar lander due to its characteristics such as light weight, high energy absorption efficiency and applicability under severe space environments. Crush strength of the honeycomb should have strength to endure during shock energy absorbing process. In this paper, the crush strength, which depends on the shape of honeycomb and impact velocity, is estimated using FEM. Ls-dyna is used for finite element analysis of the honeycomb shock absorber. The unit cells of the honeycomb shape are modeled and used for the finite element analysis. Energy absorption characteristics are decided considering several conditions such as impact velocity, foil thickness and branch angle of the honeycomb.

• Korean Journal of Materials Research
• /
• v.28 no.2
• /
• pp.75-81
• /
• 2018

An ultra-high temperature ceramic, tantalum carbide, has received much attention for its favorable characteristics: a superior melting point and chemical compatibility with carbon and other carbides. One drawback is the high temperature erosion of carbon/carbon (C/C) composites. To address this drawback, we deposited TaC on C/C with silicon carbide as an intermediate layer. Prior to the TaC deposition, the $TaCl_5-C_3H_6-H_2$ system was thermodynamically analyzed with FactSage 6.2 and compared with the $TaCl_5-CH_4-H_2$ system. The results confirmed that the $TaCl_5-C_3H_6-H_2$ system had a more realistic cost and deposition efficiency than $TaCl_5-CH_4-H_2$. A dense and uniform TaC layer was successfully deposited under conditions of Ta/C = 0.5, $1200^{\circ}C$ and 100 torr. This study verified that the thermodynamic analysis is appropriate as a guide and prerequisite for carbide deposition.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.13 no.8
• /
• pp.3726-3734
• /
• 2012

This study evaluated the behavior of pollutants based on the seasonal change by selecting the branch river's factors that influence the outflow of pollutants in Soyang lake basin. The analysis method was the factor analysis that classified the factors of the drainage area influencing the outflow of pollutants, and evaluated selected representative factors. As a result of the study, SS and T-P factors should be classified as similar factors to the storm water runoff, and the improvement of water must be strived through managing source of pollution at the time of no rain. Second, as the result of the influence from the factors, spring and winter seasons usually exert 36% influence and summer and fall exert over 90% significant influence that the improvement of water through managing source of water seems possible. At last, the prediction about delivery pollution load considering the outflow characteristic of pollutants at the drainage area based on seasonal change by regarding selected factors as independent variables is possible.