바이모달 이산정보에 대한 신뢰성해석 기법 비교

Abstract

A distribution of response usually depends on the distribution of variable. When a distribution of variable has two different modes, response also follows a distribution with two different modes. In most of reliability analysis methods, they do not care the number of modes but the type of distribution. However in actual problem, since the information is provided often with two or more modes, it is important to estimate the distributions with two or more modes. Recently some reliability analysis methods for bimodal distribution are suggested. In this paper, we review the methods, i.e., maximum entropy principle (MEP) and Akaike information criterion (AIC), and compare with Monte Carlo simulation (MCS) using mathematical example with two different modes.