• The Korean Journal of Ceramics
• /
• 제6권1호
• /
• pp.9-14
• /
• 2000

Glass transition temperature (T/sub g/) is proportional to the quadrupole splitting(Δ) of Fe(III) obtained from the /sup 57/Fe Mossbauer spectra (T/sub g/-Δ rule (1990)). The values of Δ reflect the distortion of Fe(III) atoms, which occupy the sites of network-forming atoms. Heat treatment of potassium vanadate and calcium gallate glasses at around the individual T/sub g/ causes a structural relaxation, accompanying a linear decrease of T/sub g/ and Δ values. These experimental results prove that T/sub g/ decreases with a decrease in the distortion of VO₄, GaO₄, and FeO₄tetrahedra, as the T/sub g/-Δ rule predicted.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2008년도 추계학술발표대회
• /
• pp.1332-1335
• /
• 2008

In order to various service types of real time and non-real time traffic with varying requirements are transmitted over the IEEE 802.16 standard is expected to provide quality of service(QoS) researchers have explored to provide a queue management scheme with differentiated loss guarantees for the future Internet. The sides of a packet drop rate, an each class to differential drop probability on achieving a low delay and high traffic intensity. Improved a queue management scheme to be enhanced to offer a drop probability is desired necessarily. This paper considers multiple random early detection with differential drop probability which is a slightly modified version of the Multiple-RED(Random Early Detection) model, to get the performance of the best suited, we analyzes its main control parameters (max_th, min_th, max_p) for achieving the proportional loss differentiation (PLD) model, and gives their setting guidance from the analytic approach. we propose Dynamic-multiple queue management scheme based on differential drop probability, called Dynamic-VQSDDP(Variable Queue State Differential Drop Probability)T, is proposed to overcome M-RED's shortcoming as well as supports static max_p parameter setting values for relative and each class proportional loss differentiation. M-RED is static according to the situation of the network traffic, Network environment is very dynamic situation. Therefore max_p parameter values needs to modify too to the constantly and dynamic. The verification of the guidance is shown with figuring out loss probability using a proposed algorithm under dynamic offered load and is also selection problem of optimal values of parameters for high traffic intensity and show that Dynamic-VQSDDP has the better performance in terms of packet drop rate. We also demonstrated using an ns-2 network simulation.

• 한국산업정보학회논문지
• /
• 제29권1호
• /
• pp.49-62
• /
• 2024

본 연구는 국내 모바일폰 산업을 위한 지속가능한 폐쇄루프 공급망 (Sustainable closed-loop supply chain: SCLSC) 네트워크 모델을 제안한다. 제안된 SCLSC 네트워크 모델의 지속 가능성을 위해 경제적, 환경적, 사회적 요인들이 각각 고려된다. 이들 세 가지 요인들은 SCLSC 네트워크 모델의 각 단계에서 고려되는 설비의 구축 및 운영으로부터 발생하는 총비용 최소화, CO₂ 방출 총량 최소화, 사회적 영향력 최대화를 목표로 한다. 이러한 목표들은 SCLSC 네트워크의 모델링 단계에서 각각 개별적인 목적함수로 고려되어야 하기 때문에 SCLSC 네트워크 모델은 다목적 최적화 문제로 간주할 수 있다. SCLSC 네트워크 모델은 수리모델을 사용하여 표현되며, 혼합형 메타휴리스틱 접근법을 수리모델에 적용하여 그 해를 구한다. 수치실험에서는 제안된 혼합형 메타휴리스틱 접근법의 수행도가 기존의 메타휴리스틱 접근법들의 수행도와 비교된다. 실험결과는 본 연구에서 제안된 혼합형 메타휴리스틱 접근법이 기존의 메타휴리스틱 접근법들과 비교하여 더 뛰어난 수행도를 보여주는 것을 알 수 있다.

• 한국멀티미디어학회논문지
• /
• 제23권6호
• /
• pp.747-755
• /
• 2020

In this paper, we deal with outlier data problems that occur when constructing a PM_2.5 fine dust forecasting system using a neural network. In general, when learning a neural network, some of the data are not helpful for learning, but rather disturbing. Those are called outlier data. When they are included in the training data, various problems such as overfitting occur. In building a PM_2.5 fine dust concentration forecasting system using neural network, we have found several outlier data in the training data. We, therefore, remove them, and then make learning 3 ways. Over_outlier model removes outlier data that target concentration is low, but the model forecast is high. Under_outlier model removes outliers data that target concentration is high, but the model forecast is low. All_outlier model removes both Over_outlier and Under_outlier data. We compare 3 models with a conventional outlier removal model and non-removal model. Our outlier removal model shows better performance than the others.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2014년도 춘계학술발표대회
• /
• pp.785-787
• /
• 2014

본 논문에서는 화산 분출 초저음파의 식별을 위해서 FSPNNC(Frequency Sub-bands Parallel Neural NetworkClassification)을 선택한다. FSPNNC 는 각기 다른 주파수 영역에서 독립적으로 추출한 특징벡터를 병렬 구조의 신경망에 학습하는 구조를 가지며 하나의 신경망은 하나의 분류 및 하나의 주파수 부대역만을 학습하고 다른 신경망들은 해당 특징벡터를 분류하지 않도록 학습된다. 실험은 단일 신경망 및 PNNCB(Parallel Neural Network Classifier Bank)와의 비교실험을 통하여 식별 성능을 제시한다.

• 한국안광학회지
• /
• 제7권2호
• /
• pp.27-31
• /
• 2002

4성분 $Li_2O-B_2O_3-Al_2O_3-SiO_2$ 유리들을 $R({\equiv}Li_2Omole%/B_2O_3mole%)$과 $K({\equiv}(Al_2O_3mole%+SiO_2mole%/B_2O_3mole%)$에 의해 제작하여 유리들의 구조를 굴절률 (refractive index)과 Vicker's 경도(hardness)의 변화를 측정하여 분석하였다. 먼저, 굴절률의 증가는 유리 내부구조의 분극률을 증가시키는 $Li^+$ 양이온 수의 증가에 우선적으로 의존하여 증가하였으며, 적은 양의 리튬 산화물($Li_2O$)이 첨가된 영역에서는 굴절률은 리튬 이온 양에 의존하며, 많은 양의 리튬 산화물이 첨가된 영역에서는 큰 몰 부피를 갖고 하나의 비가교 산소를 갖는 $BO_3{^-}$ 단위들의 형성으로 유리 구조 내의 몰 부피 증가로 유리들의 굴절률의 증가가 둔화되었다. 그리고 알루미늄 산화물($Al_2O_3$)과 규소 산화물($SiO_2$)의 증가에 따라 굴절률의 감소는 $Al_2O_3$와 $SiO_2$에 의해 형성되어진 $AlO_4$ 단위들과 $SiO_4{^-}$ 단위들이 붕소 산화물($B_2O_3$)에 의해 형성되어진 $BO_4$단위들보다 몰 부피의 증가로 감소되어졌다. 또한, 경도의 증가는 유리 망목구조에 형성되어지는 $BO_4$ 단위 수에 의존하였으며, 경도의 감소는 유리 망목구조를 개방화시키는 $BO_3{^-}$ 단위 수에 의존하여 감소함을 알 수 있었다.

• Korean Chemical Engineering Research
• /
• 제61권1호
• /
• pp.123-141
• /
• 2023

인공신경망을 이용한 모델 개발에서 데이터의 품질은 모델 성능에 큰 영향을 주고, 양질의 충분한 데이터가 인공신경망 훈련을 위해 필요하다. 하지만, 공학 분야에서는 적은 양의 데이터로 모델을 개발해야 하는 경우가 자주 발생한다. 본 논문은 토양에 살포된 축산 분뇨로부터 암모니아 방출량에 대한 적은 수의 데이터(83 개)를 사용하여 인공신경망 모델의 예측 성능을 향상할 수 있는 방안을 제시하였다. Michaelis-Menten 식으로 표현되는 암모니아 방출량 문제는 11개 입력변수에 대하여 2개 출력변수로 구성되었다. 출력변수는 최대 질소 발생량(N_max, kg/ha)과 N_max의 절반에 도달하는 시간(K_m, h) 이다. 범주형 입력변수에 대해 다차원 등간격 기법인 one-hot encoding 을 이용하여 데이터 전처리를 수행하였고, 훈련데이터 66개에 대하여 generative adversarial network (GAN)을 이용하여 13개 데이터를 추가로 보강하였다. 또한, 인공신경망의 초매개변수인 은닉층 수, 각 은닉층 내 뉴런 수, 활성화 함수의 최적 조합을 찾기 위하여 Gaussian process (GP)를 사용하였다. 기존의 인공신경망 구조(Lim et al., 2007) 는 17개 평가데이터에 대하여 mean absolute error (MAE)는 K_m에서 0.0668, N_max에서 0.1860이었다. 본 연구에서 제시된 인공신경망 모델은 K_m에서 0.0414, N_max에서 0.0818로 MAE 가 기존 모델 대비 각각 38%, 56% 감소하였다. 본 연구에서 제시된 방법은 적은 양의 데이터를 갖는 문제에서 인공신경망 성능을 향상하기 위하여 활용할 수 있을 것이다.

• 대한산업공학회지
• /
• 제42권5호
• /
• pp.337-351
• /
• 2016

As technologies' life-cycle shortens and their development directions are uncertain, firms' technology planning capability becomes increasingly important. Prior patent-based studies using technology growth curves identify developmental stages of technologies, thereby formulating technology development directions from an overall perspective. However, a technology generally consists of multiple sub-technologies and accordingly their development stages are likely various. In this regard, the prior studies failed to identify core sub-technologies and their specific development directions. Therefore, we suggest an approach consisting of 1) identifying core sub-technologies of a given technology using patent co-classifications and social network analysis, and 2) identifying each sub-technology's development stage and thereby determining its further development direction. We apply our approach to patents related to augmented reality to examine its applicability. It is expected that our approach will help identify evolving development stages for the core sub-technologies of a given technology, thereby effectively assisting technology experts in technology planning processes.

• Journal of Computational Design and Engineering
• /
• 제2권2호
• /
• pp.105-112
• /
• 2015

Product Lifecycle Management (PLM) ranges from design concepts of products to disposal. In this paper, we focus on the production planning phase in PLM, which is related to process planning and production scheduling and so on. In this study, key decisions for the creation of production plans are defined as production-planning attributes. Production-planning attributes correlate complexly in production-planning problems. Traditionally, the production-planning problem splits sub-problems based on experiences, because of the complexity. In addition, the orders in which to solve each sub-problem are determined by priorities between sub-problems. However, such approaches make solution space over-restricted and make it difficult to find a better solution. We have proposed a representation of combinations of alternatives in production-planning attributes by using Zero-Suppressed Binary Decision Diagrams. The ZDD represents only feasible combinations of alternatives that satisfy constraints in the production planning. Moreover, we have developed a solution search method that solves production-planning problems with ZDDs. In this paper, we propose an approach for managing solution candidates by ZDDs' network for addressing larger production-planning problems. The network can be created by linkages of ZDDs that express constraints in individual sub-problems and between sub-problems. The benefit of this approach is that it represents solution space, satisfying whole constraints in the production planning. This case study shows that the validity of the proposed approach.

• Safety and Health at Work
• /
• 제10권4호
• /
• pp.452-460
• /
• 2019

Background: Monitoring and control of PM_2.5 are being recognized as key to address health issues attributed to PM_2.5. Availability of low-cost PM_2.5 sensors made it possible to introduce a number of portable PM_2.5 monitors based on light scattering to the consumer market at an affordable price. Accuracy of light scatteringe-based PM_2.5 monitors significantly depends on the method of calibration. Static calibration curve is used as the most popular calibration method for low-cost PM_2.5 sensors particularly because of ease of application. Drawback in this approach is, however, the lack of accuracy. Methods: This study discussed the calibration of a low-cost PM_2.5-monitoring device (PMD) to improve the accuracy and reliability for practical use. The proposed method is based on construction of the PM_2.5 sensor network using Message Queuing Telemetry Transport (MQTT) protocol and web query of reference measurement data available at government-authorized PM monitoring station (GAMS) in the republic of Korea. Four machine learning (ML) algorithms such as support vector machine, k-nearest neighbors, random forest, and extreme gradient boosting were used as regression models to calibrate the PMD measurements of PM_2.5. Performance of each ML algorithm was evaluated using stratified K-fold cross-validation, and a linear regression model was used as a reference. Results: Based on the performance of ML algorithms used, regression of the output of the PMD to PM_2.5 concentrations data available from the GAMS through web query was effective. The extreme gradient boosting algorithm showed the best performance with a mean coefficient of determination (R²) of 0.78 and standard error of 5.0 ㎍/㎥, corresponding to 8% increase in R² and 12% decrease in root mean square error in comparison with the linear regression model. Minimum 100 hours of calibration period was found required to calibrate the PMD to its full capacity. Calibration method proposed poses a limitation on the location of the PMD being in the vicinity of the GAMS. As the number of the PMD participating in the sensor network increases, however, calibrated PMDs can be used as reference devices to nearby PMDs that require calibration, forming a calibration chain through MQTT protocol. Conclusions: Calibration of a low-cost PMD, which is based on construction of PM_2.5 sensor network using MQTT protocol and web query of reference measurement data available at a GAMS, significantly improves the accuracy and reliability of a PMD, thereby making practical use of the low-cost PMD possible.