Structural damage identification by extended Kalman filter withl1-norm regularization scheme
| Autor: | Chun Zhang, Lin Chen, Jie-Zhong Huang, Gu-Quan Song |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
020101 civil engineering
02 engineering and technology Building and Construction Inverse problem Regularization (mathematics) Invariant extended Kalman filter 0201 civil engineering Tikhonov regularization Extended Kalman filter 020303 mechanical engineering & transports 0203 mechanical engineering Mechanics of Materials Robustness (computer science) Control theory Fast Kalman filter Ensemble Kalman filter Civil and Structural Engineering Mathematics |
| Zdroj: | Structural Control and Health Monitoring. 24:e1999 |
| ISSN: | 1545-2255 |
| DOI: | 10.1002/stc.1999 |
| Popis: | Summary The extended Kalman filter (EKF) is a powerful tool used to assess structural damage in the time domain. A method combining EKF and l2-norm regularization, such as Tikhonov regularization, is found to improve the solution of ill-posed inverse problems. However, the l2-norm regularization process may lead to an over-smooth damage identification solution, which is contradictory to the sparse and concentrated distribution of local damages. This paper presents a new damage identification algorithm based on EKF with l1-norm regularization via free vibration responses. The l1-norm regularization item is used to enhance the identification accuracy of local damages while restraining the interference of measurement noise. Afterward, the constrained minimization problem is solved by EKF endowed with a pseudomeasurement equation. The numerical and experimental examples confirm that the proposed algorithm shows good robustness and excellent accuracy of damage identification with the unknown initial structural state. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|