Exploration of Machine Learning Approaches to Predict Pavement Performance
-
2018-03-23
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Resource Type:
-
Geographical Coverage:
-
Corporate Publisher:
-
Abstract:Machine learning (ML) techniques were used to model and predict pavement condition index (PCI) for various pavement types using a variety of input variables. The primary objective of this research was to develop and assess PCI predictive models for the years 2014 and 2015 based upon the 2013 PCI values and other road characteristics during calendar year 2013. Clearly, if a road segment was resurfaced during 2014 or 2015, then this information was expected to profoundly affect the PCI for 2015. Data collected by the Iowa Department of Transportation (DOT) regarding road conditions across the state of Iowa were used to model PCI. IBM’s Watson Analytics was utilized as a ML tool to perform the analysis. The analysis shows that ML is a viable approach to modelling PCI for various pavement types and that it is possible to predict future PCI from past PCI values, which thus eliminates the need to measure PCI for road segments on a yearly basis. This approach also has an advantage over multiple linear regression models in that it automatically accounts for nonlinear relationships.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
File Type: