Open AccessHybrid kernel identification method based on support vector regression and regularisation network algorithms
Author(s) -
Taouali Okba,
Elaissi Ilyes,
Messaoud Hassani
Publication year - 2014
Publication title -
iet signal processing
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.384
H-Index - 42
ISSN - 1751-9683
DOI - 10.1049/iet-spr.2013.0242
Subject(s) - support vector machine , computer science , kernel (algebra) , identification (biology) , algorithm , kernel method , regression , artificial intelligence , kernel regression , polynomial kernel , pattern recognition (psychology) , mathematics , statistics , botany , combinatorics , biology
This study proposes a new kernel method for online identification of a non‐linear system modelled on reproducing kernel Hilbert space (RKHS). The proposed method is a concatenation of two techniques proposed in the literature, the support vector regression and the Regularisation Networks (RNs). The proposed algorithm, called the online SVR‐RN kernel method, uses first the SVR in an offline phase to construct an RKHS model with a reduced parameter number and second the RN method in an online phase to update the model parameters. The proposed algorithm has been tested to identify the chemical Tennessee Eastman Process and the electronic non‐linear system with a Wiener Hammerstein structure.