Use of Machine Learning to Predict Long-Term Skid Resistant of Concrete Pavement

An adequate level of skid resistance over the service life of concrete pavements is crucial for the safety of drivers, especially in wet weather. It has been known that frictional properties of concrete pavements are influenced by concrete mixture proportions, type/properties of aggregates, surface texturing, and degree of surface polishing. Several experimental studies have attempted to establish regression correlations between these factors with time-dependent frictional properties of concrete pavements. While these experiments are necessary, they are costly and labor-intensive. As such, the current project intends to use the datasets and body of information generated by these past studies to develop a robust prediction algorithm for frictional properties of concrete pavements using the power of machine learning. More specifically, artificial neural network (ANN) is employed to resolve highly complicated relationships between frictional properties of concrete pavements and the parameters that influence such properties (e.g., aggregate mineralogy, concrete mixture proportions, etc.). Both the time-dependent frictional properties and terminal friction values are investigated. This report also provides a broad literature review on the subject.

Language

  • English

Media Info

  • Media Type: Digital/other
  • Edition: Final Report
  • Features: Figures; Photos; References; Tables;
  • Pagination: 50p

Subject/Index Terms

Filing Info

  • Accession Number: 01836953
  • Record Type: Publication
  • Report/Paper Numbers: CIAM-COR-R28, LTI 2022-03
  • Contract Numbers: 69A3551847103
  • Files: UTC, NTL, TRIS, ATRI, USDOT
  • Created Date: Feb 24 2022 5:18PM