ORCID

0000-0001-8013-6036

Keywords

Artificial Intelligence, Structural Health Monitoring, Machine Learning, Vehicle-to-Everything Communication, Bridge Inspection, Roadway Digital Infrastructure

Abstract

Civil transportation infrastructure systems, including highway bridges and traffic signal support structures, are primarily evaluated through periodic inspection practices based on visual assessment and limited nondestructive testing. While these methods remain the regulatory standard, they are constrained by subjectivity, limited accessibility to critical components, and the inability to capture structural condition continuously across large infrastructure inventories. These limitations are especially significant when attempting to detect visible surface deterioration or structural damage. There is a need to develop novel, practical technologies designed to complement existing inspection standards. This study addresses this need. Particular emphasis is placed on concrete bridge components and on identifying localized or concealed damage within structural connections of ancillary systems, such as traffic signal mast arms.

This dissertation formulates infrastructure condition assessment as a local damage observability problem and develops an edge-enabled framework for detecting and quantifying both visible and hidden deterioration using complementary sensing modalities. The research integrates computer vision, structural dynamics, and machine learning within deployable monitoring architectures tailored to field conditions.

The first component focuses on concrete highway bridges, where edge-deployable convolutional neural network models are developed for real-time detection and quantification of surface defects, specifically cracks and spalling. A two-stage approach combining object detection and semantic segmentation enables localization and geometric characterization of defects directly during inspection using edge devices and mixed reality platforms.

The second component addresses hidden damage in traffic signal mast-arm structures, where deterioration (corrosion, bolt loosening) occurs in non-visible components such as base anchor bolts and connections. A physics-guided machine learning framework is developed to infer stiffness reductions from vibration-based features derived from modal frequencies and intermodal relationships.

The final component extends these approaches to population-scale monitoring by integrating edge computation with connected infrastructure systems, enabling distributed sensing and continuous condition assessment without dense instrumentation.

Results demonstrate reliable surface defect quantification and accurate inference of localized hidden damage under variability and noise, supporting scalable, data-driven monitoring of transportation infrastructure systems.

Completion Date

2026

Semester

Spring

Committee Chair

Dr. Necati Catbas

Degree

Doctor of Philosophy (Ph.D.)

College

College of Engineering and Computer Science

Department

Civil, Environmental and Construction Engineering

Format

PDF

Document Type

Dissertation

Identifier

DP0053251

Release Date

5-15-2027

Download

Available for download on Saturday, May 15, 2027

Share

COinS
 

Accessibility Statement

This item was created or digitized prior to April 24, 2027, or is a reproduction of legacy media created before that date. It is preserved in its original, unmodified state specifically for research, reference, or historical recordkeeping. In accordance with the ADA Title II Final Rule, the University Libraries provides accessible versions of archival materials upon request. To request an accommodation for this item, please submit an accessibility request form.