Optimal design of MAS-ADT considering the influence of minimum accelerated stress

Abstract

The minimum acceleration stress directly affects the extrapolation accuracy and acceleration effect of the degradation model of the accelerated degradation test, which in turn affects the accuracy of the reliability assessment and the efficiency of the accelerated test. Aiming at the problem that the minimum acceleration stress is given empirically, this paper proposes a method to determine the minimum acceleration stress by considering the extrapolation accuracy and acceleration effect. The optimal minimum accelerated stress is obtained by the algorithm, and other design variables such as the sample size and the number of tests are taken into account to establish a multi-objective optimal design model of minimum accelerated stress-accelerated degradation test (MAS-ADT), genetic algorithm and gradient descent are applied to search for the optimal solution with the highest model accuracy. The sensitivity analysis of the proposed optimization model is carried out, and the results show that the model has good robustness. Finally, the optimal design of electrical connectors with and without cost constraints is investigated, which shows that the proposed method has good practical application value.