Alternative Approach for Modeling and Simulation of Pavement Deterioration Models: Artificial Neural Networks

Timely identification of undesirable cracks, raveling, rut conditions, and roughness has been a critical step in pavement management at network level. To date many models have been developed for forecasting of pavement conditions and most popular of them in developing countries is World Bank developed model HDM-4. This study summarizes the implementation of a pavement condition prediction methodology using Artificial Neural Network (ANN) for four individual ANN models to forecast cracking, raveling, rut conditions and roughness for Low Volume Roads (LVR). Road inventory data as well as six cycles of pavement performance data (pre-monsoon, post-monsoon and during winter season) including various pavement distresses, sub grade characterization and traffic data have been collected for 61 in service LVR pavement sections in plain, rolling and mountainous terrains during the years 2004, 2005 and 2006 of India and ANN models with different architectures are trained and tested using the above database to suggest the optimum architecture. Modeling results suggested ANN models in the study have a capacity to satisfactorily forecast future individual distress conditions of cracking, ravelling, rut conditions, and roughness. The performance of the suggested ANN models are compared with that of the calibrated HDM-4 models for prediction of distresses for sample LVR validation sections in the study area.

Language

  • English

Media Info

  • Media Type: CD-ROM
  • Features: Figures; References; Tables;
  • Pagination: 16p
  • Monograph Title: TRB 86th Annual Meeting Compendium of Papers CD-ROM

Subject/Index Terms

Filing Info

  • Accession Number: 01043831
  • Record Type: Publication
  • Report/Paper Numbers: 07-2244
  • Files: TRIS, TRB
  • Created Date: Feb 8 2007 6:55PM