Identification of weakly nonlinear systems based on Support Vector Machines

摘要

In this work we analyze the application of Support Vector Machines for Regression (SVRs) to the problem of identifying weakly nonlinear systems. Examples of simple linear and nonlinear systems are considered, taking into account both non-recursive and recursive models. When defining the SVR estimating function, several kinds of kernels are employed, and the effect on the accuracy performance of reducing the training set size is studied.