Optimal Change-Point Analysis of Pavement Condition Data for Identification of Homogeneous Sections

Abstract

This paper examined the applicability of the pruned exact linear time (PELT) algorithm for optimal change-point analysis to delineate pavement into homogeneous sections based on road condition. The primary objective of the study was to address some of the limitations associated with existing delineation approaches, such as the AASHTO cumulative difference approach (CDA) and the Bayesian binary segmentation method. The paper presents a mathematical framework for the optimal change-point analysis of pavement deflection data collected on a runway of an international airport. Through an analytical comparative study, the paper demonstrated how the proposed approach is better than existing delineation methods in terms of accuracy in identifying the change points, flexibility in the choice of the number of homogeneous sections, and computational time. The approach provides an opportunity for practitioners and road agencies to integrate their experience and knowledge of the road network into decision-making about the location of change points and their count without compromising computational accuracy. The proposed method was found to be robust because the solutions are independent of the parameters chosen for the analysis.