Integrate infrastructure performance monitoring using automatic crack evaluation system and convolutional neural network

The overall goal of this study is to develop a framework integrating infrastructure performance evaluation leveraging advanced evaluation system so-called automatic crack evaluation system (ACES) and advanced machine learning (ML) techniques (e.g., convolutional neural network (CNN)), which ultimately enable reliable traffic disruption-free assessment, provide structural performance data incorporating with the damage, and help accurate prediction of structural damage with proper damage classification. The proper maintenance and operation of deteriorating infrastructures require timely detection, precise diagnosis, and accurate estimation of possible structural performance degradation induced by various damages. Current technologies have been developed to prevent time-consuming and labor-intensive field tests. However, most high-speed techniques still present practical challenges (1) still have some limitations in accuracy, sensitivity, and coverage by focusing on indirect response and external surface conditions, (2) not considering structural performances that are not readily available for engineers, decision-makers, and stakeholders. To improve the current system, PI developed rapid damage inspection without interrupting traffic using an ACES with the current state bridge inspection project. The noncontact manner in the system enables faster, easier, and more accurate evaluations for improving timely maintenance. PI also has performed finite element (FE) analysis to simulate the real situation for damaged structure incorporating with damage index map obtained from ACES. Unfortunately, modeling of the damage simulation will be a challenge with tremendous efforts for each bridge, while ACES will provide quick and real-time internal damage. To address the challenges of developing both disruption-free damage inspection and effective integration of structural performance with their data results from FE simulation and ACES will be analyzed by ML technique including CNN, which is an artificial neural network that has so far been the most advanced and popularly used for analyzing images for efficient infrastructure maintenance. After fundamental validation of cracks on bridge decks with experiment, numerical simulation, and ML, further simulation and ML model study will be conducted with damages in other critical structural members to provide the comprehensive prediction models in different ML features by defects. The primary objectives of the proposal are (1) to improve nondestructive testing (NDT) systems for structural health monitoring (SHM) by using ACES with the high-speed reference-free damage detecting system and state-of-art signal processing algorithms to enhance damage recognition capability and speed, (2) to perform FE modeling for an efficient structural performance incorporating ACES data; and (3) to develop a CNN framework that provides a quick decision of its structure performance to make reliable asset management decisions. The significant outcome is to develop a highly reliable and efficient NDT system leveraging state-of-the-art prediction models for the improvements of infrastructure maintenance. If the proposed objectives and outcomes are successfully achieved, after the deep-learning training of FE results with damage location, CNN help to understand and provide a quick solution of total infrastructure evaluation effectively. Ultimately, the proposed systems will, in turn, provide an effective monitoring system and ideal asset management strategies to extend the service life of the transportation infrastructure. Thus, the study will meet transportation departments’ goals and objectives for transportation safety in urban and rural areas. The technology can be applicable for other data analysis or classification problems, as well as a specialization for being able to pick out or detect patterns by CNN. This cutting-edge neural network technique has proven its statistical power in identifying the unknown state of objects, including image classification and natural language process. The project will also carry out the Trans-SET missions by performing research, technology transfer, education, workforce development, and outreach activities to solve transportation challenges in Region 6.

• Record URL:
  https://transet.lsu.edu/wp-content/uploads/sites/16/2021/08/21CUTA01.pdf
• Supplemental Notes:
  • 21CUTA01

Language

• English

Project

• Status: Active
• Funding: $134000
• Contract Numbers:

  69A3551747106

• Sponsor Organizations:

  Office of the Assistant Secretary for Research and Technology

  University Transportation Centers Program
  Department of Transportation
  Washington, DC  United States  20590
• Managing Organizations:

  Transportation Consortium of South-Central States (Tran-SET)

  Louisiana State University
  Baton Rouge, LA  United States  70803
• Project Managers:

  Mousa, Momen

• Performing Organizations:

  University of Texas at Arlington

  Box 19308
  Arlington, TX  United States  76019-0308
• Principal Investigators:

  Ham, Suyun

• Start Date: 20210801
• Expected Completion Date: 20230201
• Actual Completion Date: 0
• USDOT Program: University Transportation Centers Program

Subject/Index Terms

• TRT Terms: Bridge engineering; Cracking; Neural networks; Nondestructive tests; Structural health monitoring
• Subject Areas: Bridges and other structures; Highways; Maintenance and Preservation;

Filing Info

• Accession Number: 01833030
• Record Type: Research project
• Source Agency: Transportation Consortium of South-Central States (Tran-SET)
• Contract Numbers: 69A3551747106
• Files: UTC, RIP
• Created Date: Jan 20 2022 1:56PM