Data analysis and reliability estimation of step-down stress accelerated degradation test based on Wiener process

According to the present works, it is more efficient by carrying out accelerated degradation test (ADT) with step-down stress than constant and step-up stress. However, maximum likelihood estimation (MLE) is currently the main analysis method for step-down stress accelerated degradation test (SDSADT), which is difficult to be applied into small sample size occasion. To solve this problem, the paper proposes a Bayesian analysis method for SDSADT based on Wiener process. Firstly, degradation path of product performance is described by Wiener process, and then data analysis of SDSADT is carried out by a segmented Bayesian method which can update the evaluation values of model parameters under different accelerated stresses. Lastly, recursive algorithm for parameters estimation is established for reliability assessment of SDSADT. Simulation studies show that the method is more accuracy by comparing with MLE in case of small sample size.