Comparison of Stochastic Response Surface Method and Monte Carlo Method for Uncertainty Analysis of Electronics Prognostics

The uncertainties in prognostics have an effect on the applicability of prognostics methods, and the quality and the degree of trustiness of prognostics results. Monte Carlo method is the most common method for uncertainty analysis. But it is a time-consuming method and the simulation time consumed improves as the sampling times improve. This will cost a large amount of computing sources. In this paper, the prognostics uncertainty analysis method based on stochastic response surface method (SRSM) has been proposed. In the case study of the board-level electronic product prognostics of a strain tester, the second order SRSM is selected for uncertainty analysis. The comparison shows that the prognostics result based on the SRSM of 27 times simulation is close to the result based on the Monte Carlo method of 100,000 times simulation. It verifies the rapid convergence and effectiveness of the SRSM for the prognostics uncertainty analysis.