Session: 08 - 01 NDE for Civil Infrastructure

Paper Number: 118548

118548 - Automated Ground Penetrating Radar Data Processing Algorithm in Reinforced Concrete Structures 

Automated Ground Penetrating Radar Data Processing Algorithm in Reinforced Concrete Structures

Lihong Mao¹, Nur Yazdani² and Eyosias Beneberu³

^1&2Department of Civil Engineering, University of Texas at Arlington, P. O. Box 19308 Arlington, TX 76019-0308, USA

³Bridgefarmer & Associates, Inc., 2350 Valley View Lane, Dallas, TX  75234, USA

ABSTRACT: Ground Penetrating Radar (GPR) is widely used for the non-destructive evaluation (NDE) of reinforced concrete (RC) structures to determine concrete cover and locate reinforcement. Existing GPR data processing methods require manual interaction, which is time-consuming and labor-intensive. This study developed a comprehensive algorithm for fully automating GPR data processing by integrating image processing techniques, a column-connection clustering (C3) algorithm, and a hyperbola-fitting machine learning model. The first step involved importing the raw data into MATLAB and interpreting the recorded data. In the second step, the time-zero correction was applied to obtain corrected B-scan images. In the third step, image processing techniques, such as moving average filtering, threshold, and C3 algorithm, were used to localize the regions that contain potential hyperbolas. Then, a machine learning model was applied to identify the hyperbolic signatures. In the last step, the peak of each hyperbola is localized, and its corresponding vertical time axis and horizontal distance axis were identified to calculate the rebar location in the concrete structure. The proposed methodology was validated using GPR scan data of 40 RC samples with varying rebar sizes, depths, and spacing. The data processing algorithm predicted the rebar depth with an average accuracy of  95%. Expected applications of the proposed methodology include: (1) Facilitating accurate and efficient assessment of the structural integrity of RC structures, such as bridges, tunnels, and buildings; and (2) Offering an efficient and reliable approach for fully automated processing of GPR data, significantly reducing labor and time for NDE of RC structures.

¹Graduate Research Assistant, E-mail: lxm6578@mavs.uta.edu

²Professor, Ph.D., P.E., F. ASCE, F.ACI, F.SEI, E-mail: yazdani@uta.edu

³ Bridge Engineer, Ph.D., PE, F.SEI, E-mail: eyosias.beneberu@mavs.uta.edu

Presenting Author: Lihong Mao The University of Texas at Arlington

Presenting Author Biography: Lihong (April) Mao is a talented Civil Engineering Ph.D. student currently enrolled at the University of Texas at Arlington. Her Ph.D. research is focused on developing an integrated algorithm for automatic Ground Penetrating Radar (GPR) data processing for concrete rebar location. Lihong has already made significant contributions to this field through her development of a Matlab application that can process raw GPR data to obtain rebar information with full automation. Her work has resulted in three published papers, including "A Novel Time-Zero Correction Method in GPR Data Processing," "Nondestructive Methods to Estimate GPR Propagation Velocity and Rebar Radius," and "Comprehensive GPR Data Processing Algorithm in Reinforced Concrete Structures."
In addition to her Ph.D. research, Lihong is also the leader of the Non-Destructive Evaluation team in her lab. In this role, she has managed and conducted safety evaluations for four TXDOT bridges and one culvert, demonstrating her proficiency in using non-destructive devices such as GPR, Impact-echo, tomography, iCOR, and Infrared camera.
With her expertise in Civil Engineering and her passion for developing innovative solutions, Lihong is a rising star in her field. Her dedication to her research and her commitment to the field of Civil Engineering have earned her recognition and accolades from her peers and mentors, and she is poised to continue making significant contributions to her field in the years ahead.