Reliability is an important attribute of a product's quality. Hence, assessing a product's reliability information is an essential task of improving continously the product's quality. Accelerated degradation tests (ADTs) are widely used to assess the reliability information for very-highly reliable products whose quality characteristics degrade over time very slowly. In estimating a product's reliability information, the interval estimation is more favorite to the manufacturer than the point estimation. Several decision variables, such as the inspection frequency, the sample size, and the termination time at each stress level, are closely related to the precision of the interval estimation and the experimental cost in an ADT. Clearly, an inappropriate setting of these decision variables wastes the experimental resources as well as reduces the precision of data analysis. The purpose of this paper is to deal with the problem of designing an ADT such that the interval estimation of the mean-time-to-failure (MTTF) at use condition is efficient. More specifically, with respect to the products whose degradation rates follow a lognormal distribution, under the constraint that the total experimental cost does not exceed a pre--determined budget, a mixed nonlinear programming is built to determine the optimal combinations of these decision variables at each stress level and the optimal combination of the CIs of the parameters involved in the MTTF's expression such that the expected width of a 100(1-γ )% CI of the MTTF is minimal.
SUMMARYStatistically designed experiments provide a proactive means for improving reliability; moreover, they can be used to design products that are robust to noise factors which are hard or impossible to control. Traditionally, failuretime data have been collected; for high-reliability products, it is unlikely that failures will occur in a reasonable testing period, so the experiment will be uninformative. An alternative, however, is to collect degradation data. Take, for example, fluorescent lamps whose light intensity decreases over time. Observation of light-intensity degradation paths, given that they are smooth, provides information about the reliability of the lamp, and does not require the lamps to fail. This paper considers experiments with such data for 'reliability improvement', as well as for 'robust reliability achievement' using Taguchi's robust design paradigm. A two-stage maximum-likelihood analysis based on a nonlinear random-effects model is proposed and illustrated with data from two experiments. One experiment considers the reliability improvement of fluorescent lamps. The other experiment focuses on robust reliability improvement of light-emitting diodes.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.