A fast polynomial chaos expansion for uncertainty quantification in stochastic electromagnetic problems

Tomy, G J K and Vinoy, K J (2019) A fast polynomial chaos expansion for uncertainty quantification in stochastic electromagnetic problems. In: IEEE Antennas and Wireless Propagation Letters, 18 (10). pp. 2120-2124.

PDF ieee-ant_wir_pro_let_18-10_2120-2124_2019.pdf - Published Version Restricted to Registered users only Download (502kB) | Request a copy

Abstract

Variations in material properties, boundary conditions, or the geometry can be expected in most electromagnetic problems. When these uncertainties in different regions of the model space are considered simultaneously, the stochastic dimensionality and the computational cost increase. Hence, uncertainty quantification in such problems is seldom attempted even though its quantification leads to a robust model. In this letter, a nonintrusive least square polynomial chaos expansion method is employed to quantify uncertainty due to stochastic variation of material properties. Using this method, the deviation from the mean performance for the transmission coefficient is obtained across the operational frequency range. The results compare well with Monte Carlo method and require just 1 of its total computational time.

Item Type:	Journal Article
Publication:	IEEE Antennas and Wireless Propagation Letters
Publisher:	Institute of Electrical and Electronics Engineers Inc.
Additional Information:	Copyright of this article belongs to Institute of Electrical and Electronics Engineers Inc.
Keywords:	Electromagnets; Finite element method; Least squares approximations; Monte Carlo methods; Polynomials; Stochastic models; Uncertainty analysis, Generalized minimal residuals; High stochastic dimensionality; Least Square; Non-intrusive; Polynomial chaos expansion (PCE), Stochastic systems
Department/Centre:	Division of Electrical Sciences > Electrical Communication Engineering
Date Deposited:	24 Jan 2020 10:46
Last Modified:	24 Jan 2020 10:46
URI:	http://eprints.iisc.ac.in/id/eprint/64376

Actions (login required)

View Item