Wiener processes with random initial degradation values for step-stress accelerated degradation tests data

Today, many products are of high reliability. It takes a long time before its failure. Accelerated degradation testing (ADT) is a common way to assess the reliability of such products. However, sufficient samples are required in constant-stress ADT for lifetime inference. It is difficult to perform the usual constant-stress ADT under the constrain of really limiting samples, like some very expensive products. In this circumstance, usually only a few test samples are available. The step-stress accelerated degradation testing (SSADT) can be used to overcome this difficulty. Due to the differences between samples, the degradation levels of different products may have random initial values. This will bring additional uncertainty for the lifetime inference of the product. Wiener process is adopted in this study to model the step-stress degradation data, considering the inconsistency of the initial degradation performance of the product. The method of parameter inference under this model is given. The effectiveness of the method is verified on specific data.