NUMERICAL ANALYSIS AND APPLICATION ON COVARIANCE-DRIVEN STOCHASTIC SUBSPACE METHOD IN MODAL PARAMETERS IDENTIFICATION

Resumo

The SSI-COV method (Covariance-driven Stochastic Subspace Identification) is a technique
within time domain modal identification methods that uses ambient vibrations as the input forces for the
identification of modal parameters (natural frequencies, damping ratios and vibration modes). In this
method two alternatives are available for the identification of the state space transition matrix: a)
applying the decomposition property of a shifted block Toeplitz correlation matrix or b) by applying the
shift property of the observability matrix. In this research, the SSI-COV was applied for the
determination of dynamic characteristics (natural frequencies and damping ratios) of a concrete block
of the Itaipú Hydroelectric Dam. This dam is equipped with a monitoring system, currently in operation,
which collects acceleration data. For the implementation, the method was programmed in the Python
language and validated through two types of simulations in which the sensitivity of the method was
evaluated. Then, for the identification of modal parameters of the concrete block, it was applied to the
acceleration records from a sensor installed in it the two alternatives for identification. Finally, the
obtained results from the two variants to compute the state transition matrix allowed us to define that
applying the shift property of the observability matrix is more advantageous in terms of data accuracy
and computational cost.