Asphalt Pavement Roughness Modeling Using the Artificial Neural Network and Linear Regression Approaches for LTPP Southern Region

The objective of this study is to develop the pavement roughness deterioration model using multiple linear regression (MLR) and Artificial Neural Network (ANN) approaches. These models consider construction number intervention (CN) factor as a dummy variable to take into account the maintenance and repair treatments on the pavement section. Data sets from 34 asphalt pavement test sections in the Long-Term Pavement Performance (LTPP) Southern U.S. states were analyzed. The data sets for asphalt pavement with bound base included in the General Pavement Study 2 (GPS-2) of the LTPP database were compiled. The independent variables are initial International Roughness Index (IRI sub 0), pavement age, equivalent single axle load (ESAL) traffic application, design structural number, as well as the construction number. In general, consideration of CN improves coefficient of determination (R²) for both MLR equations and ANN models. The MLR equations showed relatively low R² values compared to the ANN models. However, despite low R² value, the predicted IRI values using the MLR showed reasonable difference of 12.2% compared to the measured IRI values. Overall, the analysis indicated that the ANN models with high R² values outperform the MLR equations in predicting the pavement roughness. Analysis on the three different ANN analysis approaches conclude that the feedback ANN model provides the highest R² value compared to other ANN models and MLR equation. The comparison with previous study proved that the feedback ANN and the enhanced MLR equation provided accurate IRI predictions compared to the Mechanistic-Empirical Pavement Design Guide (MEPDG) equation for the Southern U.S. states. The results indicate that both ANN model and the MLR equation show small percent difference of 5.3% to 5.5% with the measured IRI values, respectively, compared to 23.1% of the MEPDG predictions.

• Supplemental Notes:
  • This paper was sponsored by TRB committee AFD20 Standing Committee on Pavement Monitoring and Evaluation. Alternate title: Asphalt Pavement Roughness Modeling Using the Artificial Neural Network and Linear Regression Approaches for LTPP Southern U.S. States.
• Corporate Authors:

  Transportation Research Board

  500 Fifth Street, NW
  Washington, DC  United States  20001
• Authors:
  • Mohamed Jaafar, Zul Fahmi Bin
  • Uddin, Waheed
  • Najjar, Yacoub
• Conference:
  • Transportation Research Board 95th Annual Meeting
  • Location: Washington DC, United States
  • Date: 2016-1-10 to 2016-1-14
• Date: 2016

Language

• English

Media Info

• Media Type: Digital/other
• Features: Figures; References; Tables;
• Pagination: 18p
• Monograph Title: TRB 95th Annual Meeting Compendium of Papers

Subject/Index Terms

• TRT Terms: Asphalt pavements; Linear regression analysis; Mathematical models; Multiple regression analysis; Neural networks; Roughness
• Identifier Terms: International Roughness Index; Long-Term Pavement Performance Program; Mechanistic-Empirical Pavement Design Guide
• Uncontrolled Terms: Coefficient of determination
• Geographic Terms: Southern States
• Subject Areas: Design; Highways; Pavements; I22: Design of Pavements, Railways and Guideways; I23: Properties of Road Surfaces;

Filing Info

• Accession Number: 01590068
• Record Type: Publication
• Report/Paper Numbers: 16-4191
• Files: TRIS, TRB, ATRI
• Created Date: Feb 9 2016 5:22PM