Model structure selection for discrete-time systems using memetic algorithm
System identification involves several steps: data acquisition, model structure selection, parameter estimation and model validation (Ljung, 1999) and research on system identification has been widely reviewed (Astrom and Eykhoff, 1971; Billings, 1980; Haber and Keviczky, 1978). One of the compon...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Book Section |
| Published: |
Penerbit UTM
2007
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/13568/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Universiti Teknologi Malaysia |
| Summary: | System identification involves several steps: data acquisition, model structure selection, parameter estimation and model validation (Ljung, 1999) and research on system identification has been widely reviewed (Astrom and Eykhoff, 1971; Billings, 1980; Haber and Keviczky, 1978). One of the components in system identification is model structure selection. In model structure selection based on regression model, expanding the system input and output lags and non-linearity functions involve excessive number of candidate terms. There are several ways to determine which terms are the significant ones to be included in the model such as stepwise regression (Billings and Voon, 1986), f-test (Söderstrom and Stoica, 1989) and orthogonal least squares (OLS) method (Billings et al., 1988). However, these algorithms require a lot of computation especially when involving large number of possible candidate terms. |
|---|