Comparing Neural Networks and Ordered Probit Models for Forecasting Pavement Condition in New Mexico
The decision making process for allocating and scheduling maintenance and rehabilitation activities within a road network depends on reliable and effective methods to effectively predict pavement’s performance over its service life. Recent studies have identified Probabilistic Neural Networks Model (PNNM) and Ordered-Probit Models (OPM) as reliable tools for predicting pavement behaviors. The objective of this study is to compare the prediction capabilities of each model in forecasting the pavement distress rate. Both models were developed using traffic data as the independent variables and pavement’s distress rate as the dependent variable. Their performance were compared from the estimation results using a dataset of 5,888 observations collected in Northern New Mexico. The findings of this study suggest that PNNM can be more accurate for predicting distress rate than OPM and can be used as input for a variety of practices in pavement management including investment, design, and rehabilitation policy. In addition, the models suggest that variables such as peak hour volume and single heavy commercial average volume are significant predictors of the distress rate for a pavement. Although both models have been proven to be effective and reliable in predicting pavement deterioration, there has not been a study comparing their performance. This study provides a better understanding in their performance when compared to each other. The application of reliable predictive models, such as the PNNM, facilitates the decision-making process.
-
Supplemental Notes:
- This paper was sponsored by TRB committee AFD10 Standing Committee on Pavement Management Systems. Alternate title: Comparing Artificial Neural Networks and Ordered-Probit Models for Forecasting Pavement Condition in New Mexico,
-
Corporate Authors:
500 Fifth Street, NW
Washington, DC United States 20001 -
Authors:
- Sanabria, Natalia
- Valentin, Vanessa
- Bogus, Susan
- Zhang, Guohui
- Kalhor, Elmira
-
Conference:
- Transportation Research Board 96th Annual Meeting
- Location: Washington DC, United States
- Date: 2017-1-8 to 2017-1-12
- Date: 2017
Language
- English
Media Info
- Media Type: Digital/other
- Features: Figures; References; Tables;
- Pagination: 15p
- Monograph Title: TRB 96th Annual Meeting Compendium of Papers
Subject/Index Terms
- TRT Terms: Decision making; Forecasting; Neural networks; Pavement distress; Pavement management systems; Pavement performance
- Uncontrolled Terms: Ordered probit models
- Geographic Terms: New Mexico
- Subject Areas: Highways; Pavements;
Filing Info
- Accession Number: 01622436
- Record Type: Publication
- Report/Paper Numbers: 17-01037
- Files: TRIS, TRB, ATRI
- Created Date: Jan 16 2017 5:06PM