Discipline: Technology and Engineering
Subcategory: Civil/Mechanical/Manufacturing Engineering
Dequante Mckoy - Saint Augustine's University
Co-Author(s): Devin Harris, University of Virginia; Mohamad Alipour, University of Virginia
In this day and age, the assessment of infrastructure and health monitoring has the potential to play a vital role in determining the degradation and the life span of infrastructures. For these infrastructure components, there are numerous techniques and methods suitable for describing operational performance and safety; however, we are quickly learning that much of the information needed can be collected in non-invasive ways such as crowdsourcing and image-based measurement. The focus of this work centers on the use of these two techniques for assessing the condition state and performance of existing infrastructure systems.
Crowdsourcing is based on a practice of obtaining information for a community of users through mobile devices such as cellphones, tablets, smart glasses and Global Positioning System (GPS). Through the processing of images of concrete surfaces and analyzing the cracks once the images are analyzed it will be possible to teach a computer how to recognize and detect the existence of cracks on concrete surfaces; a process that is called ‘Computer Vision’ by ‘Machine Learning’. A critical component of this effort is the collection and labeling of image data by humans, which is described as annotation. A computer model will later be ‘trained’ to ‘learn’ the patterns in images and the annotations they received through human vision. This type of crack detection can play a vital role in being able to manage and observe health of infrastructures.
Complementary to the crowdsourcing approach, image-based behavior measurement provides a mechanism to describe the performance of existing infrastructure. For example, monitoring the health of infrastructure can also inform engineers on the deflection of bridges as different weight loads go over the bridges. To this end, a non-contact image-based technique called Digital Image Correlation (DIC) aims at tracking the pixels in a sequence of images taken from a structure as it deflects. The application of this technique was studied in laboratory experiments as well as data from field tests on highway bridges. With this research being conducted this could help the advancements of prolonging the lifespan of infrastructures and also maintaining the health of infrastructures overall.
Funder Acknowledgement(s): University of Virginia ; VA-NC Alliance
Faculty Advisor: Devin Harris, dharris@virginia.edu
Role: Majority of the research was done by myself the literature review, testing, and also data analysis.