ABSTRACT
The step-down stress Accelerated Life Testing (ALT) can be utilized to obtain the information about reliability within feasible duration, which can meet the marketplace demands for decreased development time and cost in design and development phase. The numerical methods for Maximum Likelihood Estimators (MLEs) are often employed to estimate the model parameters of stepdown stress ALT because it is difficult to obtain the analytical solution. However, the solutions of ten depend on the initial values of model parameters. As the initial values of model parameters are not needed in the Artificial Fish-Swam Algorithm (ASFA), the ASFA is employed to solve the MLEs of the model parameters for step-down stress ALT in this paper. Two examples are illustrated to validate the proposed method for Weibull and lognormal distributions, respectively.
