Delineation of Pavement Stretches into Homogeneous Sections Using Pavement Condition Data: An Optimization Approach

Abstract

AbstractThe paper illustrates a dynamic programming-based search algorithm known as the Pruned Exact Linear Time (PELT) algorithm to produce a solution for an optimization problem to identify change-points within the measurement series to delineate a pavement stretch into homogeneous sections. The data used in this paper was measured using a Falling Weight Deflectometer (FWD) in the runway stretch of an international airport with a length of 3220 m. This paper also comprehensively demonstrates two other existing methods, namely Cumulative Difference Approach (CDA) and Bayesian Segmentation Method, for delineating the same pavement stretch into homogeneous sections. Some of the shortcomings of both the methods have been discussed and the algorithms have been modified to overcome few of the drawbacks. This paper also formulated a comparative study between the three methods in terms of their sensitivity toward type of data, profound mathematical definition of homogeneity, computational complexity and ability for post-segmentation analysis for identification of homogeneous sections.KeywordsPruned exact linear time (PELT)Cumulative difference approach (CDA)Homogeneous sections