Session Track
November 02 (Thursday)
통계기반 해석 및 설계
Oral,
제14발표장(402A호),
11:00~12:00
- 윤병동(서울대)
Th14C-2
11:00~12:00
정확한 신뢰성 해석을 위한 집단화된 입력 데이터의 추가 샘플 개수 추정 기법
Uncertainty quantification of random variable for reliability analysis is usually performed by using data with exact values. However, some experimental data contain only frequency of data in a certain interval. Such data are referred to as grouped data. Because of the high cost or the low repeatability of experiments, the obtained grouped data are used in reliability analysis without considering its statistical characteristics. The accuracy of reliability analysis using grouped data by conventional methods is lower than that using data with exact values because of lack of information. In this paper, multinomial distribution is used to model the probability of interval of grouped data and Akaike information criterion is employed to select the fittest distribution for grouped data. To increase the accuracy of reliability analysis using grouped data, this paper proposes a method that estimates the number of additional sample size required considering the probability and confidence interval of grouped input data. In order to verify the proposed method, the reliability analysis is performed by using the given grouped data and the grouped data with additional sample. The results are compared with the results of Monte Carlo simulation.
Keywords : Grouped data(집단화된 데이터), Multinomial distribution(다항 분포), Maximum likelihood estimation(최대 우량 추정), Akaike information criterion(아카이케 정보척도), Confidence interval (신뢰구간), Reliability analysis(신뢰성 해석)
Paper : Th14C-2.pdf
(사)대한기계학회, 서울시 강남구 테헤란로 7길 22 한국과학기술회관 신관 702호, Tel: (02)501-3646~3648, E-mail: ksme@ksme.or.kr