Rank-Based Optimized Sequential Sampling Approaches for Reliability Prediction Under Hybrid Degradation Process

For engineering applications, the accurate reliability prediction of long-time running electronic devices used in satellites is essential. Degradation data-based approaches have become crucial and practical for reliability in recent years. With advancements in technology, satellites can be equipped with monitors that will easily collect the degradation data for important electronic devices. Nevertheless, transmitting data from satellites to the ground is challenging due to resource limitations. The issue of reliability prediction for electronic devices based on selective real-time degradation data has been insufficiently explored in academic research. To address such issue, existing works proposed the local C-optimal approach for Gaussian processes. In fact, the degradation process of electronic devices often involves continuous changes over time and the sudden changes by transient electrical stress shocks. This study investigates this hybrid degradation process with Wiener process and Poisson shocks. Furthermore, the rank-based sequential D- and E-optimal sampling approaches of degradation data for reliability prediction are proposed. The effectiveness of the proposed approaches is demonstrated through an illustration and simulation studies based on Metal-Oxide-Semiconductor Field-Effect Transistor devices in the intelligent power distribution system of spacecraft satellites. The result of simulations shows that the proposed approaches achieve a better performance than the local C-optimal approach.