Mining Co-Occurrence Patterns among Deep Road Distresses Using Association Rule Analysis

Abstract

Co-occurrence patterns among different deep road distresses (distresses below the road surface) play a pivotal role in road maintenance. It is essential for the sustainable development of road performance and draws much attention from road maintenance departments. However, current work mainly focused on the rapid detection and development evaluation of pavement distress. Few studies shed light on the relationship among them. In this paper, over 200 km of field tests were conducted on 87 sections of the highways in China by ground-penetrating radar (GPR). Based on the distress detection results, the association rule mining algorithm Apriori was applied to explore the co-occurrence pattern among 13 types of deep road distress. Results indicate a significant correlation among light distresses (distresses with light degree), and between light distress and severe distress (distresses with moderate and heavy degree). Light distress has an average 53% probability of accompanying or inducing other distress, which is supposed to be maintained in time to prevent the road from further deterioration. Light and severe distress have a 36% probability of co-occurrence. However, the relationship among severe distresses is proved to be weak. Compared with the external environment, the interaction between different distresses is also a considerable inducement for pavement performance deterioration. The study provides a new perspective on the generation mechanism of deep road distress, which can further help the authorities optimize the maintenance schedule.