Eventos Anais de eventos
DINAME 2017
XVII International Symposium on Dynamic Problems of Mechanics
Operational modal analysis under wind load using stochastic subspace identification
Submission Author:
Gustavo Brattstroem Wagner , RJ
Co-Authors:
Gustavo Brattstroem Wagner, Damien Foiny, Rubens Sampaio, Roberta Lima
Presenter: Gustavo Brattstroem Wagner
doi://10.26678/ABCM.DINAME2017.DIN17-0193
Abstract
Operational Modal Analysis (OMA) is an identification method of dynamical systems using output-only signals. Different from the classical approach, Experimental Modal Analysis (EMA), where the input signal are also measured, OMA has only to guarantee that the inputs satisfy some properties that are usually satisfied when the inputs are random. This fact makes OMA very flexible. It allows system identification in situations that EMA cannot meet, as is the case of large and heavy structures, where a controlled input is hard to perform and expensive. With its majors developments happening in the early 1990s, the applications of the OMA in the structural dynamic is only in the beginning. In order to master OMA one has to know some stochastic tools. Nowadays, OMA has been used as tools in two main areas. The first one is in the model validation and updating connected with large structures such as bridges, tall buildings, stadiums, and oil rigs. These structures have in common the heavy weight and the ambient forces they are subjected to, as wind, traffic, and waves which are difficult to measure. Therefore, OMA methods for parameters estimations suits very well. The other application where OMA has being used and developed in the recent years is in structural health monitoring (SHM), where the stiffness reduction caused by cracks or corrosion may be monitored by the changes in the modal parameters. The purpose of this article is to show a procedure of system identification in a real structure using stochastic subspace method. Becoming popular in the 2000s, stochastic subspace identification (SSI) consists in a collection of techniques that can be formulated in a consistent mathematical framework and has its origins in control theory. The two main methods found in the literature, covariance-driven and data-driven, are given similar treatment using the system controllability matrix that can be estimated using covariance matrices or orthogonal projections of the output signal. For a better understanding of such methods, the extension of the state space model is done with the arrangement of the data in Hankel matrices, which the dimensions are to be determined from the analysis and are close connected with the estimation of covariances and the system order. The method is applied to a real structure that simulates a two-floor building, where the natural frequencies, damping factors, and modes are identified under wind excitation. Different test are performed with different conditions of fluid-structure interaction to demonstrate the method potential through results comparisons. After a good overview of the method, an analysis of the results is also shown. The main tool is a stabilization diagram of the parameters that identifies the order of the system. Another way of validate the results is using the response autocorrelation. It can be shown that the response autocorrelation function can be reconstructed from the identified parameters and matched with the autocorrelation of the measured output signal. When the system identification is performed, it is highly recommended to apply more than one method, so the results can be compared. In this article the results of other output-only modal parameters estimation methods in time domain are shown. When dealing with natural frequency and damping factor, the results comparison can be done in a straightforward way since they are numbers, but the same cannot be done for the identified modes. For them it is usual to perform a Modal Assurance Criterion (MAC), which consist in a correlation measure that informs if the modes shapes are similar or not.
Keywords
Operational Modal Analysis, System Identification, Stochastic subspace methods, Wind excitation, experimental validation
