15th Triennial World Congress of the International Federation of Automatic Control
  Barcelona, 21–26 July 2002 
FINITE ELEMENT GALERKIN MODELS AND IDENTIFIED FINITE ELEMENT MODELS - A COMPARATIVE STUDY
Stephen A. Billings, Daniel Coca
Department of Automatic Control & Systems Engineering, University of
Sheffield, Mappin Street, Sheffield, S1 3JD, UK

A new approach to derive a finite element discretisation of PDE equations soley from observations is compared, in terms of approximation accuracy, with the standard finite element Galerkin approach which assumes knowledge of the governing PDE’s. It is shown both in theory and by means of an example that, for a given model order, the identified model is more accurate than the equivalent finite element Galerkin approximation derived from the original PDE’s.
Keywords: Distributed-Parameter Systems, Finite Element Solutions, Identification Algorithms, Interpolation Approximation, System Identification
Session slot T-Fr-M02: Identification of Distributed Parameter Systems/Area code 3a : Modelling, Identification and Signal Processing