Data-based Evaluation of Regression Models for Bridge Component Deterioration
Bridge management system (BMS) recommends optimizing the allocation of Maintenance, Repair & Rehabilitation (MRR) funds for a network of bridges based on available resources and bridge future component conditions. Accurate prediction of bridge conditions over time is critical for determining a reliable MRR strategy. Based on bridge performance data collected in the past, regression models are routinely used by many researchers and state agencies to predict future bridge condition. However, these regression models can be different based on the modeling assumptions, and the selection of the best model can be quite challenging and subjective in nature. In this study, a data-based objective model evaluation procedure was developed to help end-users select among different regression models. As an example, the procedure was implemented to compare two predictive models for bridge condition ratings in the state of North Dakota. The results showed that the proposed approach can reflect the true accuracy of the prediction models and also will lead to a wide range of practical applications.
-
Supplemental Notes:
- This paper was sponsored by TRB committee AHD35 Bridge Management.
-
Corporate Authors:
500 Fifth Street, NW
Washington, DC United States 20001 -
Authors:
- Lu, Pan
- Pei, Shiling
- Tolliver, Denver
- Jin, Zhibin
-
Conference:
- Transportation Research Board 94th Annual Meeting
- Location: Washington DC, United States
- Date: 2015-1-11 to 2015-1-15
- Date: 2015
Language
- English
Media Info
- Media Type: Digital/other
- Features: Figures; References; Tables;
- Pagination: 13p
- Monograph Title: TRB 94th Annual Meeting Compendium of Papers
Subject/Index Terms
- TRT Terms: Bridge management systems; Bridges; Condition surveys; Evaluation; Regression analysis; Structural deterioration and defects
- Geographic Terms: North Dakota
- Subject Areas: Bridges and other structures; Maintenance and Preservation; I61: Equipment and Maintenance Methods;
Filing Info
- Accession Number: 01551347
- Record Type: Publication
- Report/Paper Numbers: 15-1620
- Files: TRIS, TRB, ATRI
- Created Date: Jan 27 2015 11:22AM