Reliability prediction based on Birnbaum–Saunders model and its application to smart meter

For accelerated degradation testing, data analysis based on stochastic process has been drawn much attention. However, there is significant difference in reliability prediction based on different process. This paper proposes a unified distribution model combined with a stochastic process model in multiple accelerated stress degradation test. To solve the problem of heterogeneous population of pseudo failure data, the Birnbaum–Saunders model is considered as a unified distribution model for different Gaussian family process. To give an example, a detailed proof for substituting the Birnbaum–Saunders model for the first-passage time of Wiener process is provided. Then, the influence of the parameters of the Birnbaum–Saunders model was analyzed, which provided a basis for the Birnbaum–Saunders model to be selected as a unified model. To verify presented model, a case study of Smart Meter ADT is conducted. And the obtained results of this work is compared with former work of Smart Meter ADT modeling, which verifies the effectiveness of the proposed modeling method to heterogeneous population.