ABSTRACT
Systematic reliability analyses are crucial and therefore are being regularly conducted by reliability practitioners to continuously monitor the current working status and verify the reliability of their engineered products and machines. Estimation of reliability measures and their statistical comparison among different products, conditions, environments, etc. are components in this process. Reliable estimators must be used for estimating reliability measures, and efficient hypothesis tests should be employed in this regard. It is customary to utilize a parametric procedure that incorporates an assumption regarding the probability distribution of the lifetimes concerned. The identification and suggestion of appropriate distributions for non–monotone reliability functions are cumbersome tasks. They can be seen as a major drawback in the standard parametric approach given the fact that non-monotone behaviors are more prominent than monotone models in real-life scenarios.
