Abstract
In this paper the uncertainties of modal parameters estimated using ARMA models for identification of civil engineering structures are investigated. How to initialize the predictor part of a Gauss-Newton optimization algorithm is put in focus. A backward-forecasting procedure for initialization of the predictor is proposed. This procedure is compared with a standard prediction error method optimization algorithm in a simulation study. It is found that the uncertainties can be reduced by a proper selection of the initial conditions for the predictor.
Original language | English |
---|---|
Title of host publication | Proceedings of the 13th International Modal Analysis Conference : February 13-16, 1995, Nashville, Tennessee |
Number of pages | 7 |
Volume | 2 |
Place of Publication | Bethel, Connecticut |
Publisher | Society for Experimental Mechanics |
Publication date | 1995 |
Pages | 1115-1121 |
ISBN (Print) | 0-912053-48-8 |
Publication status | Published - 1995 |
Event | The International Modal Analysis Conference - Nashville, Tennessee, United States Duration: 13 Feb 1995 → 16 Feb 1995 Conference number: 13 |
Conference
Conference | The International Modal Analysis Conference |
---|---|
Number | 13 |
Country/Territory | United States |
City | Nashville, Tennessee |
Period | 13/02/1995 → 16/02/1995 |
Series | IMAC |
---|---|
Volume | 2 |
ISSN | 1046-6770 |
Keywords
- Modal Parameters
- Auto-Regressive Moving Average Models
- ARMA Models
- Gauss Newton Algorithm