Reliability estimation based on inverse Gaussian process supported by an ANN using two types of accelerated testing data

Reliability analysis relies on data support, and artificial neural network (ANN) have clear advantages in data fitting. Therefore, ANN have been combined with inverse Gaussian process in reliability estimation. Existing reliability estimation methods based on inverse Gaussian process and ANN are only suitable for analyzing degradation test data under normal operating stress. However, to shorten the testing time, accelerated tests are widely conducted. In this study, on the basis of a generic logarithmic linear form of acceleration, the ANN-supported inverse Gaussian process is improved to a model for accelerated testing, and corresponding model training and experiment are conducted for accelerated stress relaxation degradation data and lifetime data. The experiment yielded individual degradation prediction results along with their corresponding error bands, as well as the reliability curve for the population. This confirms the effectiveness of the inverse Gaussian process-based reliability estimation method supported by ANN.