Abstract
A novel algorithm is proposed to identify time varying (TV) AR and ARMAX models. Each TV model coefficient is expanded onto a wavelet basis and the expansion parameters are estimated. An F-test and an AIC based procedure are developed to discard the insignificant terms of the expansion and provide a description for the system's trajectory as well as its order. The choice of the optimal wavelet basis for this problem is also discussed. The proposed method offers a parsimonious description of the system's trajectory, while still being able to "zoom into" the local variations where necessary.
| Original language | English (US) |
|---|---|
| Title of host publication | Conference Record of the 26th Asilomar Conference on Signals, Systems and Computers, ACSSC 1992 |
| Publisher | IEEE Computer Society |
| Pages | 125-129 |
| Number of pages | 5 |
| ISBN (Electronic) | 0818631600 |
| DOIs | |
| State | Published - 1992 |
| Externally published | Yes |
| Event | 26th Asilomar Conference on Signals, Systems and Computers, ACSSC 1992 - Pacific Grove, United States Duration: Oct 26 1992 → Oct 28 1992 |
Publication series
| Name | Conference Record - Asilomar Conference on Signals, Systems and Computers |
|---|---|
| ISSN (Print) | 1058-6393 |
Conference
| Conference | 26th Asilomar Conference on Signals, Systems and Computers, ACSSC 1992 |
|---|---|
| Country/Territory | United States |
| City | Pacific Grove |
| Period | 10/26/92 → 10/28/92 |
Bibliographical note
Funding Information:This work was supported by the National Science Foundation Grant NSF-MIP 9210230.
Publisher Copyright:
© 1992 IEEE.