Remarks on Direct Deterministic Integration to Compute Probability Distributions in EMC

In this work, a non-intrusive, deterministic method to compute the probability density function of an interesting quantity that depends on several volatile parameters is studied. The underlying model may be linear or nonlinear, and the parameters need not be statistically independent. The method relies on an approximation of the probability measure of the interesting quantity via a partition of the parameter space and integration of the joint multivariate probability density function of the volatile parameters. As an example, the method is applied to the computation of the transfer impedance of certain shielded cables. A faster asymptotic convergence as in the Monte Carlo case can be empirically shown for the considered cases.