http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
풍력 Flange 소재(S355NL)에 대한 피로수명에 관한 연구
신상훈(Sanghun Shin),김남용(Namyong Kim),이승갑(Seunggap Lee),이현우(Hyunwoo Lee) 대한기계학회 2010 대한기계학회 춘추학술대회 Vol.2010 No.11
S355NL is used in wind power structural steel. It is also improving high strength, low temperature toughness and corrosion resistance. In this study, mechanical properties were evaluated by tensile test at room temperature, -20℃ and -40℃, respectively. Based on the tensile test data, high and low cycle fatigue test were conducted on S355NL at room temperature and -20℃. Fatigue test is performed under stress-controlled conditions for the stress ratio of 0.1 and 0.2. Fatigue life predictions of the estimated linear fit are compared and discussed. As a result, it is shown that the room temperature and stress ratio of 0.2 is longer than -20℃ and that of 0.1.
Stereoscopic Imaging System 및 딥러닝을 활용한 충돌 회피 알고리즘 개발
신상훈(Sanghun Shin),이용훈(Yonghun Lee),최웅철(Woongchul Choi) 한국자동차공학회 2019 한국자동차공학회 학술대회 및 전시회 Vol.2019 No.11
Due to the development of technology, research on autonomous vehicle driving is being actively conducted. Many people want to take autonomous driving, but the technology is unreliable and high in cost and limited use. In order to increase the reliability of autonomous driving using autonomous driving sub-devices using low-cost cameras that do not use expensive Radar or LiDAR, new technologies will be discussed. With the development of deep learning technology, an image recognition algorithm called YOLO has been developed, which shows the improvement of image recognition speed and high accuracy which is a problem of the existing image recognition. In addition, it is possible to develop an algorithm to obtain accurate position, velocity and acceleration by recognizing the object measured by YOLO using stereo image matching technology. Both technologies will enable the development of low cost autonomous driving sub-equipment using cameras.
비재무 정보를 활용한 IPO 주식의 상장일 가격 등락 예측에 관한 연구
신상훈(Sanghun Shin),이현준(Hyun Jun Lee),안재준(Jae Joon Ahn) 한국데이터정보과학회 2018 한국데이터정보과학회지 Vol.29 No.2
공모주는 비상장기업이 유가증권시장 또는 코스닥시장에 상장하기 위해 불특정 다수를 대상으로 판매하는 주식을 의미하며, 이러한 과정을 기업공개라고 한다. 공모주는 판매 가격인 공모가에 대비하여 상장일 종가는 대부분 상승하는 경향을 보이며, 그 수익률이 높기 때문에 저금리 시대에 좋은 투자 대안으로 자리하고 있다. 그러나 기관투자자에 비해 개인투자자들은 공모 및 공모주에 대한 정보를 얻고 분석하기 힘들고, 따라서 실제로 성공적인 공모주 투자를 진행하기 어려운 실정이다. 따라서 본 논문은 개인투자자들이 비교적 수집하기 쉬운 비재무적 자료를 활용하여 공무가 대비 상장일 종가의 상승과 하락을 예측할 수 있는지 확인하며, 다양한 분석 방법론을 활용하여 그 정확도를 비교한다. 본 연구에 적용된 분석 방법론으로는 로지스틱 회귀분석, 판별분석, 의사결정나무, 인공신경망, 사례기반추론, 그리고 서포트벡터머신을 사용하였으며, 2007년부터 2017년 9월 6일까지의 공모주 데이터를 활용하여 실증분석을 진행하였다. The stock in a public offering refers to stocks that are sold to an unspecified public to be listed on the KOSPI or the KOSDAQ market. Because the initial public offering (IPO)’s closing price of offering day mostly tends to rise opposed to the offering price of it, which is selling price, it is a good alternative investment asset in the low interest rate period. However, it is difficult for individual investors to obtain and analyze information on public offerings and IPO stocks compared to institutional investors. Therefore, this paper confirms whether individual investors can predict the rise and fall of the IPO’s closing price of offering day compared to its offering price by using multiple data analysis methodologies and non-financial data that is relatively easy to collect, and compares the accuracy of them. Logistic regression, discriminant analysis, decision tree, artificial neural network, case based reasoning, and support vector machine were used for analysis, and the empirical experiments was conducted using the IPO data from 2007 to September 6, 2017.