A Self-Organizing Quasi-Linear ARX RBFN Model for Nonlinear Dynamical Systems Identification

Abstract

The quasi-linear ARX radial basis function network (RBFN) model has shown good approximation ability and usefulness in nonlinear system identification and control. It has an easy-to-use structure, good generalization and strong tolerance to input noise. In this paper, we propose a self-organizing quasi-linear ARX RBFN (QARX-RBFN) model by introducing a self-organizing scheme to the quasi-linear ARX RBFN model. Based on the active firing rate and the mutual information of RBF nodes, the RBF nodes in the quasi-linear ARX RBFN model can be added or removed, so as to automatically optimize the structure of the quasi-linear ARX RBFN model for a given system. This significantly improves the performance of the model. Numerical simulations on both identification and control of nonlinear dynamical system confirm the effectiveness of the proposed self-organizing QARX-RBFN model.

Nonlinear System Identification Using Quasi-ARX RBFN Models with a Parameter-Classified Scheme

“…In the last two decades, nonlinear models such as neural networks (NNs), radial basis function networks (RBFNs), neurofuzzy networks (NFNs), and multiagent networks have received considerable research attention for nonlinear system identification [1][2][3][4]. However, from a user's point of view, the conventional nonlinear black-box models have been criticized mostly for not being user-friendly: (1) they neglect some good properties of the successful linear black-box modeling, such as the linear structure and simplicity [5,6]; (2) an easy-to-use model is to interpret properties of nonlinear dynamics rather than being treated as vehicles for adjusting fit to the data [7]. Therefore, careful modeling is needed for a model structure favorable to certain applications.…”