Combining Kriging and Controlled Stratification to Identify Extreme Levels of Electromagnetic Interference

• Published in: 2019 International Symposium on Electromagnetic Compatibility - EMC EUROPE pp. 404 - 409
• Main Authors: Houret, T., Besnier, P., Vauchamp, S., Pouliguen, P.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2019
• Subjects: Computational modeling; Controlled Stratification; Electromagnetic compatibility; Estimation; Extreme Quantiles Estimation; Kriging; Mathematical model; Monte Carlo; Monte Carlo methods; Predictive models; Uncertainty; Uncertainty propagation
• ISSN: 2325-0364
• DOI: 10.1109/EMCEurope.2019.8872042

EMC risk analysis requires various configurations of coupling paths described by important sets of unknown or uncertain parameters. More specifically, values at risk corresponding to extreme values of relevant fields, currents or voltages are often the most important information with regard to a possible EMC risk. Therefore, we aim at estimating extreme quantiles of the relevant field, current or voltage. Controlled stratification accelerates the standard Empirical estimation convergence to sample output extreme values, thus reducing the required number of calls to cost-expensive full-wave simulations. However, controlled stratification requires a simple (i.e. fast calculation time) model with sufficient correlation to the initial model. The main idea in this communication is to use a surrogate model as a simple model. Kriging was previously identified as a surrogate model with relevant properties. In this paper, we investigate the performance of combined kriging and control stratification. We show that this combination outperforms the stand-alone kriging surrogate model for estimating extreme quantiles. On the contrary, the latter performs better to identify less extreme quantiles.