| Status | 已發表Published |
| A Novel Online Algorithm for Simultaneous Data Cleansing and Robust Damage Detection | |
Yuen, K. V.  ; Mu, H.Q.
| |
| 2014-12-01 | |
| Source Publication | Proceedings of 5th Asia-Pacific Workshop on Structural Health Monitoring Conference (AWPSHM 2014)
 |
| Abstract | In recent years, there has been rapid development on data acquisition system for structural health monitoring. Given the fact that data acquisition system operates under varying operational and environmental conditions, outliers may occur due to unknown disturbance and/or malfunction of instruments. Since the presence of outliers degrades the performance of structural identification techniques, data cleansing such as noise removal and outlier removal using signal processing techniques is important for structural health monitoring. In this paper, a robust extended Kalman filter-based algorithm is proposed for simultaneous data cleansing and robust structural identification using outlier-contaminated dynamic response data. It provides a rigorous solution to parametric identification and uncertainty quantification. In this algorithm, the probability of outlier, which is a function of the data size and its normalized residual, is used to evaluate the degree of outlierness of the measurement at each time step. The normalized residual is defined as the difference between the measured value and the corresponding one-step-ahead predictor, normalized by its standard deviation. The data points with probability of outlier over 0.5 will be regarded as suspicious data points and they will be discarded for identification purpose. In contrast to other existing outlier detection criteria that require subjective threshold (e.g., normalized residual larger than 2.5), the outlier probability threshold of 0.5 is intuitive in the proposed approach. Finally, the proposed algorithm will be applied to estimate the parameters of a structural system with degrading stiffness, and it turns out that the proposed algorithm is successful in simultaneously detecting outlier and capturing the degrading stiffness trend with more stable results than the plain Kalman filter. |
| Keyword | Bayesian damage detection data cleansing outlier real-time identification structural health monitoring |
| Language | 英語English |
| The Source to Article | PB_Publication |
| PUB ID | 27853 |
| Document Type | Conference paper |
| Collection | GRADUATE SCHOOL |
| Corresponding Author | Yuen, K. V. |
| Recommended Citation GB/T 7714 | Yuen, K. V.,Mu, H.Q.. A Novel Online Algorithm for Simultaneous Data Cleansing and Robust Damage Detection[C], 2014. |
| APA | Yuen, K. V.., & Mu, H.Q. (2014). A Novel Online Algorithm for Simultaneous Data Cleansing and Robust Damage Detection. Proceedings of 5th Asia-Pacific Workshop on Structural Health Monitoring Conference (AWPSHM 2014). |
| Files in This Item: | There are no files associated with this item. | |||||
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.
Edit Comment