A multivariate system reliability estimation method based on step stress accelerated degradation testing

In this paper, a multivariate system reliability estimation method based on step stress accelerated degradation testing is proposed. Firstly, the method of step stress accelerated degradation testing (SSADT) is utilized to evaluate the reliability of LED luminaries. In the test, temperature is determined as the sensitive stress. Secondly, we suppose that all the performance metrics follow normal distribution and the variation of distribution parameters with time is employed to indicate the performance degradation. Arrhenius model is used to do the extrapolation. Thirdly, the correlation coefficients between different performance metrics are described as a Dunnett-Sobel class correlation matrix to consider the statistical dependence. At the same time, the dependent performance metrics are transformed to conditional-independent random variables. Then, the failure criteria are determined according to the standard from Energy Star. Subsequently, the system reliability is calculated based on total probability theorem and Monte Carlo simulation. At last, the system reliability of the LED luminaries considering lumen maintenance, CCT and color maintenance simultaneously is obtained.