On the Identification of Frequencies and Damping Ratios for Structural Health Monitoring Using Autoregressive Models
2012 23rd International Workshop on Database and Expert Systems Applications (DEXA) (2012)
Sept. 3, 2012 to Sept. 7, 2012
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/DEXA.2012.19
Modal parameter identification plays an important role within damage identification strategies in the field of structural health monitoring. The identification of natural frequencies and damping ratios by means of dynamic measurements provides a good information basis for further analysis. In this paper we review the Dynamic Data System (DDS) approach using autoregressive (AR) models in order to overcome the limitations of the FFT-based methods and evaluate it using experimental data from a real analysis case. As a secondary problem, we also discuss an ARMA model order identification technique which will be used to determine an upper bound for the used AR models. Our results show that this model order is too low for the identification of almost every eigen frequency of the unfiltered measurement signature. Furthermore, the k-means clustering algorithm was used to clean up the data as well as to get the correct eigen frequency-damping ratio pair in a semi-automated way.
autoregressive moving average processes, condition monitoring, damping, fast Fourier transforms, parameter estimation, pattern clustering, structural engineering computing
H. Kosorus, M. Hollrigl-Binder, H. Allmer and J. Kung, "On the Identification of Frequencies and Damping Ratios for Structural Health Monitoring Using Autoregressive Models," 2012 23rd International Workshop on Database and Expert Systems Applications(DEXA), Vienna, Austria Austria, 2012, pp. 23-27.