Date of Award

Spring 5-3-2024

Level of Access Assigned by Author

Open-Access Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Civil Engineering

Advisor

Eric N. Landis

Second Committee Member

Wilhelm “Alex” Friess

Third Committee Member

Bill Davids

Additional Committee Members

Salimeh Yasaei Sekeh

Tzuyang Yu

Abstract

The aging and deterioration of civil infrastructures, particularly bridges, pose significant safety risks and economical burdens. Despite gradual improvements in bridge conditions over time, a substantial portion still requires preventative maintenance to ensure public safety. Traditional bridge inspection methods are often time-consuming, expensive, subject to human error, and can pose safety risks to both inspection crews and the public. This thesis addresses these challenges by proposing innovative solutions through the integration of advanced technologies, with a focus on unmanned aerial vehicles (UAVs), deep learning (DL) algorithms, and image fusion techniques. This research aims to enhance bridge inspection practices by improving data collection, analysis, and interpretation steps of inspection.

It begins by evaluating the impact of UAV hardware configurations on bridge inspection mission capabilities. Through this evaluation, the study highlights the suitability of different types of UAVs for various inspection tasks. Notably, quadcopter configurations are found to be optimal for visual camera inspections, while larger multicopters are better suited for inspections requiring higher payload capacities, such as hyperspectral and LiDAR sensors.

Furthermore, the study explores the application of DL algorithms for processing large volumes of inspection images to automate defect detection. The findings demonstrate promising results, showcasing the effectiveness of DL algorithms in processing images effectively and identifying various types of defects in bridge structures. This advancement holds potential for significantly accelerating the inspection process while improving accuracy.

Additionally, the thesis introduces a DL-based approach for steel bridge corrosion condition rating. Accompanied by the development of an open-source dataset for model training, this approach offers practical solutions for engineers by providing accurate rating of corrosion conditions according to bridge inspectors reference manual (BIRM) and American association of state highway and transportation officials (AASHTO) regulations.

Moreover, the efficacy of fused infrared and visible image for delamination detection in concrete bridge elements is thoroughly investigated. While fusing IR and visible images show promising results in enhancing delamination detection, the study also identifies challenges such as overlooking subsurface delaminations or detecting them with lower confidence scores in fused images compared to infrared images alone. The findings emphasize the importance of carefully considering potential sources of misinterpretation when analyzing fused and IR images for delamination detection purposes.

Download

Files over 10MB may be slow to open. For best results, right-click and select "save as..."

Share