Evolving MIMO Flexible Neural Trees for Nonlinear System Identification.
Yuehui Chen, Feng Chen, Jack Y. Yang · 2007
Abstract — The problem of identification of a nonlinear dynamic system by using multiple-input and multiple-output flexible neural tree (MIMO-FNT) is presented in this paper. This work is an extension of our previously multiple-input and singleoutput FNT model. FNT is a tree-structured neural networks which allows input variables selection, over-layer connections and different activation functions for different nodes. Based on the pre-defined instruction set, a FNT model can be created and evolved. In this research, the MIMO FNT structure is developed using the Immune Programming (IP) and the free parameters embedded in the neural tree are optimized by Particle Swarm Optimization (PSO) algorithm. Empirical results on nonlinear dynamical system identification problems indicate that the proposed method is effective and efficient. I.