Structural Health Assessment using System Identification Techniques in the Time and Frequency Domain

Abstract

Techniques used in Structural Health Assessment (SHA) are assessed. The dynamic cases of SHA are looked at thoroughly and can be further separated into time domain and frequency domain-based approaches. Frequency domain approaches are defined as approaches that compare modal parameters for indications of damage. These modal parameters include, modal frequency, damping ratio, and mode shape. Time domain-based approaches are defined as approaches that utilize time history of responses to identify structural parameters such as stiffness and damping. These two approaches each contain advantages and disadvantages over the other. In the case of the frequency domain-based approach, it is relatively quick and requires less information in order to identify the parameters of a structure. However, it only can identify damage at the global level and cannot identify damage at the local level. Time Domain based approaches can identify parameters at the local level but require a tremendous amount of data. In order to show this comparison more accurately, two methods were selected that effectively show the advantages and disadvantages of each technique. For the Time-domain approach the Unscented Kalman Filter with Unknown Input and Weighted Global Iteration (UKF-UI-WGI) was selected. It is an effective method developed at the University of Arizona by Dr. Abdullah Abdulamir Al-Hussein in 2015. The method was used to find the stiffness of each member of varying frames, and any change in the identified stiffness was taken as an indication of damage. This method was able to use a limited number of Dynamic Degrees of Freedom regardless of the frames size. However, it still requires a relatively large amount of information in order to identify the structure. For the Frequency domain-based approach, the Multivariate Auto-Regressive model with white noise Excitation was selected (MAR). This method was developed by C.W. Kim, M. Kawatani and J. Hao in 2010. This method was utilized to detect damage in Bridge structures. The method requires time-response information be collected at certain nodes along the bridge, where it is then used to identify the modal parameters of the structure. These modal parameters are then tested for their sensitivity in identifying damage types along the bridge when the bridge is excited by a moving vehicle load. This method was able to identify damage along with severity by comparing the modal parameters of each mode together. This method is unable to identify structural damage at the local level and can only tell if the structure is damaged.