SPR-2322: Artificial Intelligence (AI) and Markov Process Based Data Mining on Predicting Bridge Operating Conditions

Bridge Management Systems (BMS) has been an important tool for state agencies managing bridge networks to formulate maintenance and inspection programs within cost limitations. Prediction of future bridge condition ratings is a key and fundamental component of the BMS. A reliable bridge management program depends on how accurately the ratings of the bridge elements can be estimated. This project is aimed to develop an artificial intelligence (AI) based data mining method for estimating bridge future conditions and the deterioration process. The proposed method will also identify the impact of the most influencing parameters on the degradation of bridge structures. There are three major objectives for this research: (1) Build upon previous research to extend and refine the Artificial Neural Networks (ANN) model to provide more accurate bridge condition predictions; (2) integrate ANN with Markov process to achieve more reliable probabilistic prediction; and (3) facilitate bridge asset management and bridge inspection process in the Bridge Management System (BMS). Firstly, the proposed research will apply artificial neural networks, an important AI technique to a larger database over a longer duration i.e., 15 ~ 20 years of bridge inspection in the State of Connecticut. It is expected that these additional data extending the time range of inspections will help produce more accurate results and enable the introduction of more complicated predictions. A series of experiments with different types of neural networks and settings will be used to ensure the optimal setup is being used for the dataset being analyzed. Secondly, the research will integrate ANN results with Markov process to create deterioration curves. Markov chain modeling has been used in predicting bridge condition due to its ability to model stochastic behavior of the deterioration process. Clustering data of ANN with respect to significant influencing variables such as era of construction and percentage of heavy vehicles will help in getting more reliable results. Thirdly, the results and insights obtained through this study will be used in the bridge assessment management and inspection process, and to enable more effective and efficient design and management of bridges in the BMS. The scope of research includes the use of artificial neural networks to predict bridge operating conditions and the results will be incorporated with a Markov process to model the deterioration of bridges. The models developed will be validated using a full set of condition data collected by the Connecticut Department of Transportation (CT DOT). The predictions will be focused on the overall and level performance of bridges including deck, superstructure and substructure components. Culvert condition and channel protection prediction are not part of this research. The work presented here is considered a data driven approach for bridge condition and deterioration modelling, and mechanism-based approaches are beyond the scope of the work presented here.


    • English


    • Status: Active
    • Funding: $238,510
    • Contract Numbers:


    • Sponsor Organizations:

      Connecticut Department of Transportation

      2800 Berlin Turnpike
      P. O. Box 317546
      Newington, CT  United States  06131-7546
    • Managing Organizations:

      Connecticut Department of Transportation

      2800 Berlin Turnpike
      P. O. Box 317546
      Newington, CT  United States  06131-7546
    • Project Managers:

      Zimyeski, Melanie

    • Performing Organizations:

      University of Hartford

      200 Bloomfield Avenue
      West Hartford, CT  United States  06117
    • Principal Investigators:

      Fang, Clara

    • Start Date: 20220601
    • Expected Completion Date: 20240731
    • Actual Completion Date: 0

    Subject/Index Terms

    Filing Info

    • Accession Number: 01853898
    • Record Type: Research project
    • Source Agency: Connecticut Department of Transportation
    • Contract Numbers: SPR-2322
    • Files: RIP, STATEDOT
    • Created Date: Aug 4 2022 12:02AM