Multi-Fidelity Approach for Polynomial Chaos Based Statistical Analysis of Microwave Networks

摘要

In this paper, a novel polynomial chaos based approach for the fast statistical analysis of complex microwave structures is proposed. This approach leverages a highly efficient closed form low-fidelity model elicited from the high-dimensional model representation (HDMR) of the network. By cross-cutting the efficiency of this low-fidelity model with the accuracy of general high-fidelity simulations, the accuracy-CPU cost tradeoff for the statistical analysis can be achieved.