Online identification RKPCA-RN
Nadia Souilem, Ilyes Elaissi, Okba Taouali, Hassani Messaoud · 2013
This paper proposes a new method for online identification of a nonlinear system using Reproducing Kernel Hilbert Space (RKHS) models. The RKHS model is a linear combination of kernel functions applied to the used training set observations. For large datasets, this kernel based to severs computational problems and makes identification techniques unsuitable to the online case. For instance, in the Kernel Principal Component Analysis (KPCA) scheme the Gram matrix order grows with the number of training observations and its eigen decomposition. The proposed method is based on Reduced Kernel Principal Component Analysis technique (RKPCA), to extract the principal component will be time consuming.