Structural Damage Detection Using Compressive Sensing
Departments: Civil Engineering Advisor: Shamim Pakzad
On August 1, 2007, a gusset plate failed in the main span of the I-35W bridge in Minneapolis, Minnesota causing the entire bridge to collapse, injuring over 100 people and killing 13. It is due to catastrophes such as this that the development of structural health monitoring (SHM) systems is so important. The capability of sensing technology has improved dramatically in recent years. The latest wave of innovation has produced sensors that can gather data nearly continuously. The emergence of dense sensor networks introduces both new possibilities and new problems in data acquisition and processing. Dense sensor networks are useful in monitoring important structural components because they are able to monitor small scale damage such as fatigue cracks, crack initiation, stress concentrations, and other structural damage which is helpful for detecting damage in its early stages. However, because of the large amount of data being collected, the use of dense sensor networks becomes extremely costly if implemented without data compression techniques.
This study has consisted of the development and implementation of an iterative compressive sensing and sampling scheme for SHM. This sampling scheme in combination with a digital image correlation (DIC) system was designed to use spatial correlation to efficiently detect and localize damage on a structure. For this study, a simplified gusset plate specimen was designed, and computer modeling and simulation of the plate placed in axial tension was performed. The iterative compressive sampling scheme used a random peak detection algorithm and damage index calculations from in-plane strain data to pinpoint damage on a structure. The results of the implementation of the sampling scheme on the simulation strain data verified its effectiveness. Further testing performed with a physical specimen using a DIC system confirmed the viability of the use of the sensor and sampling setup on full scale structures. Ultimately this setup could lead to the development of reduced data intensity and cost-effective dense sensor networks that will greatly improve the ability to detect structural damage before failure occurs.
About Jamie Hudson:
Jamie Hudson, a senior at Lehigh University is pursuing a Bachelor of Science degree in civil engineering with a focus on structures and a minor in architectural history. Jamie has worked with Professor Pakzad since the end of her junior year performing research on structural damage detection using compressive sensing. Aside from research, Jamie is a member of two engineering honor societies, Tau Beta Pi and Chi Epsilon, as well as the Lehigh chapter of the American Society of Civil Engineers and the Society of Women Engineers. She is also president of Lehigh University’s Steel Bridge Team. Jamie will be attending graduate school in the fall of 2015 to pursue her masters and PhD in structural engineering.