Service-life Pavement Performance Simulation: Capturing the Influence of Traffic Flows by Big Data Analysis

Abstract

An ideal road management system requires a real-time assessment of pavement performance. But there are many root causes of the status quo that the expectation and the system have not been achieved yet. One of them is the monitoring system. In the Netherlands, the measurement is carried out once a year because of the budget and also because it is time-costly. To set up a real-time assessment system of pavement condition, this thesis proposes to use the real-time traffic data as the indirect way of monitoring the road, and applies three performance models to verify the correlation between traffic flows and road performance. The applied models are the regression models, the survival model, and the decision tree classifier. As a result, according to the test data of A15 in the Netherlands from 2015 to 2018, all the models indicate that the traffic characteristics influenced road performance, particularly, the traffic flow could worsen the pavement condition which already performed badly. The model result confirms the feasibility of establishing a real-time assessment tool of pavement condition by using the traffic data as the monitoring method. The specific establishment process of the tool is designed in the final.