Abstract
This paper aims to improve the accuracy of reliability estimates and the failure time prediction for products exhibiting nonlinear degradation behavior under constant-stress accelerated degradation test (CSADT). To achieve this, a novel degradation model and a life prediction method are proposed, which are based on a generalized Wiener process. Some models assume that the drift coefficients are related to accelerated stress. However, in certain applications, the diffusion coefficients are also affected by accelerated stress. The relationship between the drift parameter and accelerated stress variables can be derived by the accelerated model, and so is the relationship between the diffusion parameter and stress variables based on the principle of invariance of the acceleration factor. To account for individual variability among products, random effects are introduced. Model parameters are estimated using a combination of maximum likelihood estimation (MLE) and the expectation-maximization (EM) algorithm. Furthermore, the probability density function (PDF) of the remaining useful life under normal stress conditions is derived using the law of total probability. The effectiveness and applicability of the proposed approach are validated using simulated constant stress accelerated degradation data and stress relaxation data. The results demonstrate that the model not only fits the degradation process well but also modestly improves the accuracy of the failure time prediction, providing valuable guidance for engineering maintenance and reliability management.