SYSTEM IDENTIFICATION OF DYNAMIC PROPERTIES OF A TALL TOWER BY COVARIANCE DRIVEN STOCHASTIC SUBSPACE IDENTIFICATION TECHNIQUE

Abstract

Dynamic properties of structure such as natural frequency, mode shape and damping ratio are important parameters for understanding of structural responses under loads.  They can be identified by analyzing dynamic responses measured from the structure.  Identification of these parameters in civil engineering structures has several constraints due to the small number of sensors, low power vibration responses, and the input excitation is unknown and always random in nature.  Therefore, an efficient identification scheme is required for accurate results.  The objective of this research is to apply the Covariance Driven Stochastic Subspace (SSI-COV) Identification technique to improve an accuracy of dynamic property identification of structures.  This technique has been developed as an efficient tool for system identification technique.  However, the accuracy of the result is mainly dependent on selection of the size of an operating matrix called Toeplitz Matrix.  SSI-COV method would fail to yield accurate results if the size of the Toeplitz Matrix is too small.  On the contrary, if the size of the Toeplitz Matrix was set too large, the obtained results would be contaminated with some false modes and lead to difficulty in the identification.  This study proposes a technique to yield the accurate result automatically by analysis of data using different matrix sizes and the solution is determined from the weighted average based on the frequency of occurrence of the result.  Two examples are presented; simulated responses of 5 degree-of-freedom system and vibration responses from continuous measurement of a slender cylindrical shaped tower.  The advantages of SSI-COV technique are; accurate and stable results, higher mode identification, and damping identification.  The results also demonstrate the potential of an application of SSI-COV technique for an automatic identification of continuous measurement which could be useful for long-term structural health monitoring.