• 한국정밀공학회지
• /
• 제16권6호
• /
• pp.100-107
• /
• 1999

This paper describes a probabilistic remaining life assessment program for the creep crack growth. The probabilistic life assessment program is developed to increase the reliability of life assessment. The probabilistic life assessment involves some uncertainties, such as, initial crack size, material properties, and loading condition, and a triangle distribution function is used for random variable generation. The resulting information provides the engineer with an assessment of the probability of structural failure as a function of operating time given the uncertainties in the input data. This study forms basis of the probabilistic life assessment technique and will be extended to other damage mechanisms.

• 한국강구조학회 논문집
• /
• 제10권1호통권34호
• /
• pp.37-45
• /
• 1998

기계 및 구조물의 안전성과 신뢰성 평가를 위하여 금속재료의 피로강도 데이터베이스 시스템을 구축하고, 이를 이용하여 금속재료의 피로수명 분포특성을 검토하였다. 종속도를 이용하여 구성한 관계형 피로강도 데이터베이스 시스템은 일련의 데이터 관리 및 해석작업을 효율적으로 수행하였다. 또한 Robust 방법과 보상오차함수(complementary error function)를 이용한 3-파라미터 Weibull분포의 파라미터 추정법 및 비파손확률를 검토하기 위하여 제안된 새로운 기준식의 타당성을 검토하였다. 조합분포모델을 이용하여 부하응력비와 Weibull분포의 파라미터 항으로 나타낸 피로수명 분포식은 탄소강의 P-R-N 특성을 해석함에 있어서 좋은 결과를 보였다.

• 한국산학기술학회논문지
• /
• 제7권3호
• /
• pp.393-397
• /
• 2006

신제품에 대한 개발 주기가 짧아지고 있는 현시점에서 제품의 수명을 예측하고 평가하기 위한 방법으로 가속 수명시험과 시험을 통해 관측된 고장 데이터의 분석에 대한 관심이 증대되고 있다. 이에 따라 가속 수명시험을 위한 시험 조건과 고장 데이터의 정확한 분석을 위한 고장 데이터의 최적 분포 결정 방법에 대한 관심 또한 증대되고 있다. 따라서 본 논문에서는 고장 데이터를 기반으로 신뢰성 예측을 할 때 사용하는 분포 함수 결정을 위한 방법으로 관측된 고장 데이터만의 분포를 고려하는 Anderson-Darling 방법과 관측된 고장 데이터의 수명-스트레스 관계식을 적용하여 고장 데이터의 분포를 결정하는 Likelihood Function 방법을 비교한다. 두 가지 방식을 비교한 결과 각 방식에 의해 선택되는 최적분포가 다르며, 따라서 각 방식에 의해 선택된 최적 분포에 의해서 예측되는 수명도 다름을 알 수 있다.

• International Journal of Reliability and Applications
• /
• 제15권2호
• /
• pp.125-150
• /
• 2014

Accelerated life tests (ALTs) are extensively used to determine the reliability of a product in a short period of time. Test units are subject to elevated stresses which yield quick failures. ALT can be carried out using constant-stress, step-stress, progressive-stress, cyclic-stress or random-stress loading and their various combinations. An ALT with linearly increasing stress is ramp-stress test. Much of the previous work on planning ALTs has focused on constant-stress, step-stress, ramp-stress schemes and their various combinations where the stress is generally increased. This paper presents an optimal design of ramp soak-stress ALT model which is based on the principle of Thermal cycling. Thermal cycling involves applying high and low temperatures repeatedly over time. The optimal plan consists in finding out relevant experimental variables, namely, stress rates and stress rate change points, by minimizing variance of reliability function with pre-specified mission time under normal operating conditions. The Burr type XII life distribution and time-censored data have been used for the purpose. Burr type XII life distribution has been found appropriate for accelerated life testing experiments. The method developed has been explained using a numerical example and sensitivity analysis carried out.

• Journal of the Korean Data and Information Science Society
• /
• 제28권4호
• /
• pp.875-888
• /
• 2017

The Rayleigh distribution has been commonly used in life time testing studies of the probability of surviving until mission time. We focus on a reliability function of the Rayleigh distribution and deal with prior distribution on R(t). This paper is an effort to obtain Bayes estimators of rayleigh distribution with three different prior distribution on the reliability function; a noninformative prior, uniform prior and inverse gamma prior. We have found the Bayes estimator and predictive density function of a future observation y with each prior distribution. We compare the performance of the Bayes estimators under different sample size and in simulation study. We also derive the most plausible region, prediction intervals for a future observation.

• International Journal of Reliability and Applications
• /
• 제7권1호
• /
• pp.13-25
• /
• 2006

A new class of life distributions is studied. This class is defined based on comparing the residual life time to the whole life in the moment generating function order giving 'the new better than used in the moment generating function order ageing class $(NBU_{mgf})$'. Fundamental properties of this class are given including some closure properties and characterizations. Finally, we consider new results about comparisons of age and block replacement policies when the underlying distribution belongs to $NBU_{mgf}$ aging classes.

• 한국추진공학회지
• /
• 제16권4호
• /
• pp.33-41
• /
• 2012

KM6 단기추진제의 저장수명을 감마과정 이론을 이용한 확률론적 방법으로 추산하였다. 장기 저장에 따른 안정제 함량의 저하량이 0.8%일 때를 상태고장으로 보았으며 정상감마과정으로 가정하였을 때 형상함수의 상수와 척도모수를 모멘트법으로 추정하였다. 저장기간별 확률밀도함수로부터 각 저장기간에서의 상태분포를 확인할 수 있으며 누적고장분포함수 곡선에서 누적고장확률이 10%인 $B_{10}$수명은 25년이며 $B_{50}$수명은 36년으로 추산되었다. 실용적 관점에서 볼 때 $B_{50}$수명을 평균저장수명으로 볼 수 있으며 확률과정론을 이용하면 저장수명을 분포곡선으로 표현할 수 있다.

• 한국강구조학회 논문집
• /
• 제14권2호
• /
• pp.365-373
• /
• 2002

이 논문에서는 교량의 수명을 예측하기 위한 시스템 신뢰성 이론이 설명되고, 생애 분포 함수를 이용하여 현존하는 교량의 잔존 수명을 예측하는 방법이 설명된다. 시스템 이론과 생애함수 (survivor functions) 를 이용하여 LIFETIME 이라는 프로그램을 개발하였다. Survivor functions은 주어진 시간 t에 대해 신뢰성을 산출한다. 이 프로그램을 이용하여 콜로라도주에 있는 교량의 수명을 예측하였다. 이 교량은 직렬과 병렬로 구성된 시스템으로 컴퓨터 모델링 되었으며 이 모델을 이용하여 시스템 파괴 확률을 시간에 대해 계산하였다.

• 한국기계가공학회지
• /
• 제19권6호
• /
• pp.67-72
• /
• 2020

Life tests are essential in reducing the catastrophic damage caused by the accidents of large machinery such as aircraft and ships. However, life tests are challenging to implement due to the high costs and time required to test the life of large machinery parts. Therefore, it is advantageous and convenient to apply accelerated life test techniques for key components to reduce costs and time. In fact, extensive research has already been conducted on these techniques. However, recently, there have been cases in which an experimental value was applied to the shape parameter of the Weibull distribution used in the reliability test, but the test time was not significantly reduced. Therefore, in this paper, the shape parameters are estimated from the probability density function of the Weibull distribution for the analysis of an accelerated life test for bladder accumulators, which are core components of military helicopters. The test time was derived based on the number of samples and confidence level by substituting it into the test time equation. Next, the accelerated life test time was calculated using the steady-state test time with an acceleration factor obtained from the Arrhenius model. The steady-state life test required approximately 15,000 H, whereas the accelerated life test time for one sample at 100 ℃ was 34% shorter than that of the steady-state life test.

• Journal of Information Technology Applications and Management
• /
• 제24권2호
• /
• pp.71-80
• /
• 2017

Software reliability is one of the most elementary and important problems in software development In order to find the software failure occurrence, the instantaneous failure rate function in the Poisson process can have a constant, incremental or decreasing tendency independently of the failure time. In this study, we compared the reliability performance of the software reliability model using the parameters of Pareto life distribution with the intensity decreasing pattern and the shape parameter of Erlang life distribution with the intensity increasing and decreasing pattern in the software product testing. In order to identify the software failure environment, the parametric estimation was applied to the maximum likelihood estimation method. Therefore, in this paper, we compare and evaluate software reliability by applying software failure time data. The reliability of the Erlang and Pareto life models is shown to be higher than that of the Pareto lifetime distribution model when the shape parameter is higher and the Erlang model is more reliable when the shape parameter is higher. Through this study, the software design department will be able to help the software design by applying various life distribution and shape parameters, and providing basic knowledge using software failure analysis.