A cloud model-based method for the analysis of accelerated life test data

Various curve fitting models, including the Arrhenius stress model, inverse power law model, and Eyring model have been used to model the load (stress) - life relationship to aid in planning accelerated life tests; that is, the relationship between the mean of the sample lifetimes and the testing stress level. The load-life relationship is a one-to-one relationship: one mean of the sample lifetimes corresponds to one testing stress level. However, due to the random uncertainties existing in the testing stress, the relationship should be a many-to-many relationship rather than one testing stress corresponding one mean lifetime of the tested product. Based on the one-to-one relationship of the mean of the sample lifetimes to the testing stress level, a many-to-many relationship can be derived using the reasoning method presented in this paper. The reasoning method is constructed as 'If X, then Y.' X is termed the rule antecedent, and Y is called the rule consequent. They are constructed with the stress values and the sample lifetimes, respectively, based on the cloud model, which represents random uncertainty and fuzzy uncertainty. The reasoning method presented is called the multi-rule-based cloud reasoner, which can refine the one-to-one relationship established by models such as the Arrhenius stress model to a many-to-many relationship. In the case study, the multi-rule-based cloud reasoner was applied to a thermal stress accelerated life test of ammunition fuses. The results from the multi-rule-based cloud reasoner were compared with the estimation results from a normal cloud generator under a stress level of 20 °C. The results showed that the many-to-many relationship between the uncertain stress level and the means of the sample lifetimes was derived by the multi-rule-based cloud reasoner.