Polynomial Chaos Kriging Metamodel for Automotive EMC Simulations

A Polynomial Chaos Kriging metamodel is proposed for surrogate modeling and sensitivity analysis purposes of electromagnetic simulators in EMC context. To replace the costly forward model in case of a frequency dependent output, a proper metamodel is to be built many times, having an overall simulation cost that is usually not feasible with the given resources. The present paper aims at overcoming this problem by considering the frequency as a variable and using a Polynomial Chaos Kriging metamodel to handle strong frequency dependency of the model output, that often occurs when resonant electromagnetic problems are dealt with. Using an electromagnetic interference filter together with its automotive measurement setup as a test example, we illustrate the advantages and robustness of the proposed modeling approach.