• Composites Research
• /
• 제36권5호
• /
• pp.329-334
• /
• 2023

기존의 멀티스케일 유한요소법(Multiscale finite element, FE² )은 거시 스케일의 모든 적분점에서 대표 체적요소(representative volume element, RVE)의 미시 경계치 문제를 반복적으로 계산하기 때문에 긴 해석 시간과 많은 데이터 저장 공간을 필요로 한다. 이를 해결하기 위해 본 연구에서 평균장 균질화 데이터 기반 멀티스케일 해석 기법을 개발하였다. 데이터 기반 전산역학(data-driven computational mechanics, DDCM) 해석은 변형률-응력 데이터 셋을 직접적으로 사용하는 모델-프리(model-free)접근 방식이다. 멀티스케일 해석을 수행하기 위해, 평균장 균질화(mean-field homogenization)를 활용하여 복합재의 미세구조에 대한 변형률-응력 데이터베이스(database)를 효율적으로 구축하고, 이를 기반으로 데이터 기반 전산역학 시뮬레이션을 수행하였다. 본 논문에서는 개발한 멀티 스케일 해석 프레임워크(framework)를 예제에 적용하여, 초탄성(hyperelasticity) 복합재의 미세 구조를 고려한 데이터 기반 전산역학 시뮬레이션 결과를 확인하였다. 따라서, 데이터 기반 전산역학 접근 방식을 활용한 멀티스케일 해석기법은 다양한 재료 및 구조에 적용될 수 있으며, 멀티스케일 해석 연구 및 응용 가능성을 열어줄 것으로 기대된다.

• KIEAE Journal
• /
• 제17권5호
• /
• pp.77-86
• /
• 2017

Purpose: Because of the growing concern over fossil fuel use and increasing demand for greenhouse gas emission reduction since the 1990s, the building energy analysis field has produced various types of methods, which are being applied more often and broadly than ever. A lot of research products have been actively proposed in the area of the building energy simulation for over 50 years around the world. However, in the last 20 years, there have been only a few research cases where the trend of building energy analysis is examined, estimated or compared. This research aims to investigate a trend of the building energy analysis by focusing on methodology and characteristics of each method. Method: The research papers addressing the building energy analysis are classified into two types of method: engineering analysis and algorithm estimation. Especially, EPG(Energy Performance Gap), which is the limit both for the existing engineering method and the single algorithm-based estimation method, results from comparing data of two different levels- in other words, real time data and simulation data. Result: When one or more ensemble algorithms are used, more accurate estimations of energy consumption and performance are produced, and thereby improving the problem of energy performance gap.

• 동의신경정신과학회지
• /
• 제32권4호
• /
• pp.303-320
• /
• 2021

Objectives: Although medical typologies based on indigenous biopsychological ideas have been described, their integrity has been questioned due to its theory-driven nature in categorization. Therefore, studies on the Sasang typology, a temperament-based traditional Korean medicine, are needed to examine whether it is possible to classify types of specific biopsychological profiles using data-driven analysis. Methods: Psychological measures of the Eastern Sasang Personality Questionnaire (SPQ) and Western NEO-Personality Inventory (NEO-PI) along with physical measures and Sasang types were acquired from 2,049 participants. Latent groups based on the SPQ and NEO-PI subscale scores were extracted using Latent Profile Analysis. Their psychosomatic features were then compared with those of Sasang types. Results: Three SPQ-based latent groups showed distinctive psychological and physical features consistent with those of Sasang types. However, four NEOPI-based latent groups presented only psychological features. Furthermore, SPQ-High and SPQ-Low latent groups demonstrated similar psychosomatic profiles to those of So-Yang and So-Eum Sasang types, respectively. Conclusions: This study illustrates that biopsychological profiles of Sasang types are supported by psychosomatic features of latent groups based on SPQ of Eastern psychology, signifying that the categorization of Sasang typology have acceptable validity and reliability.

• 지식경영연구
• /
• 제17권1호
• /
• pp.49-72
• /
• 2016

Extant research on firm innovation productivity is limited in measuring the innovation productivity, in which they measured firm innovation productivity by using either inputs or outputs of innovation. The present study complemented the extant research by employing Data Envelopment Analysis (DEA) approach to measure firm innovation productivity. Furthermore, this paper examined the effects of firms' external knowledge search, as one of open innovation practices, on firm innovation productivity, for open innovation activities are regarded as an influencing factor on firm innovation productivity in the previous literatures. Using the data of the Korean Innovation Survey (KIS) of manufacturing industries conducted in 2008, this study developed hypotheses in which we considered not only two dimensions of external knowledge search (breadth and depth) but also two subtypes of external knowledge search (market-driven and science-driven). The results found that searching deeply and market-driven search are positively related to firm innovation productivity, but science-driven search is somewhat negatively related to firm innovation productivity. Furthermore, market-driven search can mitigate the negative effect of science-driven search on innovation productivity.

• 한국CDE학회논문집
• /
• 제22권1호
• /
• pp.80-88
• /
• 2017

Product data analytics (PDA) is a data-driven analysis method that uses product data management (PDM) databases as its operational data. It aims to understand and evaluate product development processes indirectly through the analysis of product data from the PDM databases. To educate and train PDA efficiently, this study proposed an approach that employs courses for both product development and PDA in a class. The participant group for product development provides a PDM database as a result of their product development activities, and the other group for PDA analyses the PDM database and provides analysis result to the product development group who can explain causes of the result. The collaboration between the two groups can enhance the efficiency of the education and training course on PDA. This study also includes an application example of the approach to a graduate class on PDA and discussion of its result.

• 인터넷정보학회논문지
• /
• 제19권1호
• /
• pp.123-130
• /
• 2018

With the emergence of a new paradigm called Open Science and Big Data, the need for data sharing and collaboration is also emerging in the computational science field. This paper, we analyzed data-driven research cases for computational science by field; material design, bioinformatics, high energy physics. We also studied the characteristics of the computational science data and the data management issues. To manage computational science data effectively it is required to have data quality management, increased data reliability, flexibility to support a variety of data types, and tools for analysis and linkage to the computing infrastructure. In addition, we analyzed trends of platform technology for efficient sharing and management of computational science data. The main contribution of this paper is to review the various computational science data repositories and related platform technologies to analyze the characteristics of computational science data and the problems of data management, and to present design considerations for building a future computational science data platform.

• Asia pacific journal of information systems
• /
• 제32권3호
• /
• pp.477-495
• /
• 2022

This paper proposes how to design and implement data-driven strategies by investigating how a firm can increase its value using data science. Drawing on prior studies on architectural innovation, a behavioral theory of the firm, and the knowledge-based view of the firm as well as the analysis of field observations, the paper shows how data science is abused in dealing with meso-level data while it is underused in using macro-level and alternative data to accomplish machine-human teaming and risk management. The implications help us understand why some firms are better at drawing value from intangibles such as data, data-science capabilities, and routines and how to evaluate such capabilities.

• 한국빅데이터학회지
• /
• 제8권1호
• /
• pp.35-47
• /
• 2023

It is crucial to develop effective and efficient big data analytics methods for problem-solving in the field of business in order to improve the performance of data analytics and reduce costs and risks in the analysis of customer data. In this study, a big data-driven data analysis system using artificial intelligence techniques is designed to increase the accuracy of big data analytics along with the rapid growth of the field of data science. We present a key direction for big data analysis systems through missing value imputation, outlier detection, feature extraction, utilization of explainable artificial intelligence techniques, and exploratory data analysis. Our objective is not only to develop big data analysis techniques with complex structures of business data but also to bridge the gap between the theoretical ideas in artificial intelligence methods and the analysis of real-world data in the field of business.

• 정보처리학회논문지D
• /
• 제15D권5호
• /
• pp.681-690
• /
• 2008

단백질체학에서 특정 조건 하에서 단백질의 기능 이상 및 구조 변형 유무를 규명하고 질병 과정을 추적하는 것은 중요한 연구이다. 일반적으로 단백질의 발현량 변화 분석에는 통계적 방법이 많이 사용되고 있으며 단백질 상용 이미지 분석 소프트웨어에서 제공하는 그래픽을 이용한 방법들도 있으나, 이 방법들은 많은 조직 내에 존재하는 수많은 단백질을 수동으로 비교해야 하는 어려움이 있다. 본 논문에서는 데이터베이스와 데이터마이닝 기법을 이용하여 OLAP 데이터 큐브와 Discovery-driven 탐색의 응용 방법을 제안한다. 데이터 큐브의 특성을 이용함에 의해서, 질병에 의해 발현량이 변하는 단백질 뿐 아니라 임상적 특성과 단백질의 영향 관계를 분석하는 것이 가능하다. 데이터 큐브에서 단백질의 발현량 변화 분석에 적합한 데이터 큐브의 척도와Discovery-driven 탐색을 위한 예외 지표를 제안하고, 특히 In-exception을 계산하는데 있어서의 계산량 감소 방안을 제시한다. 실험을 통해 폐암 2-DE 데이터에서 데이터 큐브와 Discovery-driven 방법이 유용함을 보인다.

• 한국시뮬레이션학회논문지
• /
• 제18권4호
• /
• pp.185-195
• /
• 2009

최근 시스템들이 복잡해지면서 시뮬레이션을 통한 시스템의 분석이 주목을 받고 있다. 시뮬레이션 분석에서 가장 핵심적인 부분 중의 하나가 시뮬레이션 모델의 검증이며, 이 과정을 통하여 시뮬레이션 모델이 얼마나 실제 시스템을 대변할 수 있는지를 판단한다. 모델의 검증에서 시뮬레이션 모델과 실제시스템의 데이터를 비교할 때 발생하는 차이는 입력 데이터의 차이에 의한 영향도 있으며, 이를 통한 모델의 검증 결과는 높은 신뢰성을 보장하지 못한다. 따라서 이 논문에서는 실제와 동일한 입력 데이터를 바탕으로 하는 Trace-Driven 시뮬레이션을 기반으로 모델을 설계하였다. 한편, 출력데이터들을 하나의 통계량을 통한 검증이 아닌 클래스 별 검증을 하기 위해 데이터마이닝 분야에서 분류기의 성능을 판단하는 F 척도를 응용하여 시뮬레이션 모델의 검증을 수행하였다. 그 결과, 제안된 검증 방법은 정밀한 모델의 검증을 가능하게 하고, 검증 시에 피드백을 제공함으로써 용이한 수정 작업을 가능하게 한다.