Tian, Na, Zhu, Longchao and Lai, Choi-Hong ORCID: https://orcid.org/0000-0002-7558-6398 (2015) Identification of boundary shape using a hybrid approach. International Journal of Machine Learning and Cybernetics, 6 (3). pp. 385-397. ISSN 1868-8071 (Print), 1868-808X (Online) (doi:10.1007/s13042-014-0266-9)

Full text not available from this repository.

Abstract

In this paper, a hybrid approach combining quantum-behaved particle swarm optimization (QPSO) and conjugate gradient method is proposed to identify boundary shape of the geometry under steady state conditions. No prior information about the shape is available, so the inverse problem is classified as function estimation. Least square method is used to model the inverse problem, which intends to minimize the difference between measured and calculated data. Considering ill-posedness of the inverse problem, Tikhonov regularization method is used to stabilize the solution. The numerical results show that the proposed hybrid method is able to recover the boundary shape, and can sharply reduce the required computation time. While considering the oscillations at the both boundaries of the estimated results, the parallel QPSO is used in order to both obtain better estimation and reduce computation time.

Item Type:	Article
Uncontrolled Keywords:	boundary element method, shape identification, quantum-behaved particle swarm optimization, Tikhonov regularization, hybrid approach
Subjects:	Q Science > QA Mathematics
Faculty / School / Research Centre / Research Group:	Faculty of Liberal Arts & Sciences
Last Modified:	17 Oct 2016 13:04
URI:	http://gala.gre.ac.uk/id/eprint/13650

Actions (login required)

View Item