Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 109
PROCEEDINGS OF THE FOURTH INTERNATIONAL CONFERENCE ON SOFT COMPUTING TECHNOLOGY IN CIVIL, STRUCTURAL AND ENVIRONMENTAL ENGINEERING
Edited by: Y. Tsompanakis, J. Kruis and B.H.V. Topping
Paper 24

Estimation of Data-Driven Polynomial Chaos using Hybrid Evolution Strategies

M.D. Spiridonakos, V.K. Dertimanis and E.N. Chatzi

Chair of Structural Mechanics, Institute of Structural Engineering, ETH Zurich, Switzerland

Full Bibliographic Reference for this paper
M.D. Spiridonakos, V.K. Dertimanis, E.N. Chatzi, "Estimation of Data-Driven Polynomial Chaos using Hybrid Evolution Strategies", in Y. Tsompanakis, J. Kruis, B.H.V. Topping, (Editors), "Proceedings of the Fourth International Conference on Soft Computing Technology in Civil, Structural and Environmental Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 24, 2015. doi:10.4203/ccp.109.24
Keywords: polynomial chaos, uncertainty, B-splines, hybrid optimization, separable non-linear least squares.

Summary
This study investigates a non-intrusive uncertainty quantification scheme relying on a polynomial chaos (PC) basis constructed from the available data. Instead of predefining basis functions with respect to the statistics of the uncertain input, the method introduced here realizes a proper basis function parametrization that adapts to a given input-output data set. Parameter estimation is effectively dealt with through a separable non-linear least squares procedure that allows for simultaneous estimation of both the PC basis and the corresponding coefficients of projection. The induced constrained non-quadratic optimization problem is treated through the implementation of a hybrid optimization algorithm that combines the advantages of evolutionary and deterministic schemes, combining higher convergence rates and reliability in the search for a global optimum. The method's effectiveness is examined through both numerical and experimental case studies, while comparisons with the classical PC based on the Wiener-Askey scheme and alternative data-driven methods, such as the arbitrary PC, provide indications of superior performance.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description