Comparison of stochastic reduced order model and stochastic collocation methods for the sensitivity analysis of crosstalk

Sensitivity analysis is a crucial tool to reduce the randomness dimension of stochastic problems, thus to reduce computation cost. However, the efficacy of different stochastic methods for sensitivity analysis was totally missing. Therefore, two popular stochastic methods, i.e., the stochastic reduced order model (SROM) and stochastic collocation (SC) methods, are performed to analyse crosstalk sensitivity to different cable variables. The Monte-Carlo (MC) method is used as the reference. Both the SROM and SC methods are shown to be much more efficient by orders of magnitude compared with the MC method. The result also shows that the SROM performance is not as remarkable as that of the SC method in terms of efficiency. As a result, the SC method could be recommended as a promising efficient approach for sensitivity analysis in stochastic electromagnetic compatibility (EMC) problems.