Genetic algorithm with an improved fitness function for (N)ARX modelling

Author(s)
Author(s)
Chen, Q.
Worden, K.
Peng, P.
Date Issued
2007
Publisher
Elsevier
Journal
Mechanical Systems and Signal Processing
Volume
21
Issue
2
Start page
994
End page
1007
Abstract
In this article a new fitness function is introduced in an attempt to improve the quality of the auto-regressive with exogenous inputs (ARX) model using a genetic algorithm (GA). The GA is employed to identify the coefficients and the number of time lags of the models of dynamic systems with the new fitness function which is based on the prediction error and the correlation functions between the prediction error and the input and output signals. The new fitness function provides the GA with a better performance in the evolution process. Two examples of the ARX modelling of a linear and a non-linear (NARX) simulated dynamic system are examined using the proposed fitness function.
URI
DOI
SFU Affiliated Publication
No
Availability at SFU Library

No database links found.