ABSTRACT
Understanding airfield pavement deterioration is essential for airport asset management to ensure safe and efficient airport operations. This paper employs Gradient Boosting Machine (GBM) – a machine learning method – to investigate the contributions of a variety of influencing factors to runway and taxiway pavement deterioration at Chicago O’Hare International Airport. By adopting a systematic procedure consisting of model training, validation, and testing, two separate GBM models are developed to estimate Pavement Condition Index (PCI) of runways and taxiways. The models account for various input variables that are believed to affect pavement deterioration, including pavement age and material, maintenance and rehabilitation history, weather conditions, and air traffic loading effects. The developed GBM models are shown to outperform other methods (including linear regression, nonlinear regression, artificial neural networks, and random forest) in terms of model goodness-of-fit for both runway and taxiway pavements. The GBM modelling results are subsequently used to interpret the influence of individual input variables as well as their interactions on PCI, using relative importance and partial dependence plots. With promising results, the study demonstrates the use of an approach that was not previously considered in infrastructure management and can help airport agencies enhance the ability to understand airport asset performance.
Acknowledgement
This research was funded by the Chicago Department of Aviation (CDA) and the O’Hare BPC/CarePlus programme through the Center of Excellence for Airport Technology (CEAT) at the University of Illinois at Urbana-Champaign. We are very grateful for the enthusiastic support from Professor David Lange, Ross Anderson, Jim Chilton, Tim Morgan, Zachary Bergman, Adam Hardy, Brad McMullen, Michael Vonic, Donald McGady, Shawn Gould, Matthew Yohn, and Carlos Mendez of this effort.
Disclosure statement
No potential conflict of interest was reported by the authors.