Fast Stochastic Surrogate Modeling via Rational Polynomial Chaos Expansions and Principal Component Analysis

This paper introduces a fast stochastic surrogate modeling technique for the frequency-domain responses of linear and passive electrical and electromagnetic systems based on polynomial chaos expansion (PCE) and principal component analysis (PCA). A rational PCE model provides high accuracy, whereas the PCA allows compressing the model, leading to a reduced number of coefficients to estimate and thereby improving the overall training efficiency. Furthermore, the PCA compression is shown to provide additional accuracy improvements thanks to its intrinsic regularization properties. The effectiveness of the proposed method is illustrated by means of several application examples.