Abstract:

To evaluate the reliability of the product with nonlinear degradation in step stress accelerated degradation test (SSADT), a performance degradation modeling and evaluation method is presented on the basis of microscopic flaw emergence and change process. Firstly, the step stress stochastic degradation model combining one-step state transition model with the Gaussian process is created to model the generation and development of micro-defects and characterize their effects to the macroscopic properties. Then, a reliability model is determined through computing the curvilinear boundary crossing probability of the Gaussian process by the optimal linear approximation. The SSADT for the dielectric resonator oscillator (DRO) is used to verify this method. The modificatory Gauss-Newton method and Powell conjugate gradient algorithm are used to get the optimal solution of the model parameters. The results indicate the state transition stochastic model based on this Gaussian process can describe performance degradation tendency and randomness, and the reasonable evaluated results could be obtained by the methodology.