Design and Evaluation Method for Small-Sample Accelerated Life Test
A small-sample accelerated life test design and evaluation method is proposed to overcome the limitations of conventional acceler-ated life tests,such as being time-consuming and requiring large samples.By conducting life tests at only two accelerated stress levels,this method can evaluate the product's reliable life at normal stress level with high confidence and high reliability.Firstly,the relationship between the lower confidence limit of the reliable life at normal stress level and the confidence limits of the reliable life at two accelerated stress levels is modeled by using the linear acceleration models(e.g.,Arrhenius model,inverse power law model and exponential model,etc.),and the method of evaluating the lower confidence limit of the reliable life at normal stress level with high confidence and high reliability is presented.Then,the accelerated life tests with exponential,Weibull and lognormal distributions and Type-Ⅱ censoring are discussed in detail,and the reliability assessment formulas are derived respectively.The method is easy to implement and suitable for complete and incomplete data as well as various distributions.The Monte Carlo simulations demonstrate that the method can significantly save the specimens and time,and is more rigorous and precise than the conventional methods which need to conduct life tests at multiple stress levels.
Accelerated life testSmall sampleTest designReliability assessmentLife predictionLower confidence limit
NETL
NSTL
万方数据
维普
