Two Applications of Eng-genes based Nonlinear Identification
| Authors: | Irwin George, Queen's University Belfast, United Kingdom
Connally Patrick, Queen's University Belfast, United Kingdom
Li Kang, Queen's University Belfast, United Kingdom |
|---|
| Topic: | 3.2 Cognition and Control ( AI, Fuzzy, Neuro, Evolut.Comp.) |
|---|
| Session: | Soft Computing for Control |
|---|
| Keywords: | Neural networks, neural network models, nonlinear systems, system identification |
|---|
Abstract
Nonlinear identification using a novel neural network paradigm, namely eng-genes, is investigated. A set of MATLAB functions for the training and simulation of eng-genes based neural models are described. These functions are then used to investigate the effectiveness of the technique applied to two nonlinear dynamical systems. Experimental data from a pH neutralisation plant and simulation data from a physical model of a CSTR process are used to generate ‘eng-genes’ models. The results are compared with conventional neural models of these plants, showing that simple neural models with better performance and improved transparency are obtainable using the eng-genes paradigm.
