Effect of Damping, Duration, and Signal-to-Noise Ratio on the Accuracy of Data-Driven Stochastic Subspace Identification

Author

Upama Bhusal, Southern Illinois University CarbondaleFollow

Date of Award

8-1-2024

Degree Name

Master of Science

Department

Civil Engineering

First Advisor

Tezcan, Jale

Abstract

The identification of modal parameters is crucial for dynamic analysis, vibration control and structural health monitoring of civil engineering structures. But, due to the size and complexity of the structures, it is quite difficult for civil engineers to get the input forces. Also, it is quite unfeasible and uneconomic to do the laboratory test of such huge structures by applying the artificial forces. So, in order to help determine the modal parameters in the cases where external forces are often unknown, various output-only identification methods have been developed. Over the past two decades, the Stochastic Subspace Identification (SSI) method has emerged as a powerful technique for modal identification using output-only data. By integrating system identification theory, linear algebra, and statistics, the SSI method can simultaneously determine natural frequencies, modal shapes, and damping ratios of multiple modes efficiently. Where civil engineers often face challenges in obtaining input data, a method like SSI is proving to be invaluable for analyzing and designing civil engineering structures. The SSI method’s computational efficiency and ability to handle closely spaced modes makes it particularly attractive for modal parameter identification compared as to other approaches.In this study, Data-driven Stochastic Subspace Identification (SSI-DATA) is used to determine the modal parameters of a five-story building idealized as a spring-mass-damper system. The study is carried out to study how well the method performs in determining the modal parameters under different levels of damping, duration, and measurement noise. MATLAB is used for all the simulations and the identified parameters are compared with the true values to investigate the effects of varying levels of measurement noise, duration, and damping using the SSI-DATA method.