Simulation-based constructions of reliability confidence intervals from degradation data

This paper focuses on the construction confidence intervals and confidence limits of products' reliability based on the distribution-based model, a widely applied degradation model. The model is briefly reviewed in the first place with a focus on the sources of uncertainties associated with it. By simulating the propagation of these uncertainties, a simulation-based approach for the constructions of reliability confidence intervals and confidence limits is proposed and discussed in this paper. Simulation studies are performed to illustrate the proposed approach and also investigate the effects of sample sizes on the precision of reliability inferences. The result shows that the precision will be increased with the growth of sample sizes until a threshold of the sample size is reached and after that, the growth of sample sizes seems to have little influence on the inference accuracy. Finally, experimental data from an accelerometer are analyzed with the proposed approach to further demonstrate its applicability.