Highway performance prediction model of International Roughness Index based on panel data analysis in subtropical monsoon climate

Abstract

Pavement performance prediction is the crucial basis for decision-making of Pavement Management Systems (PMS). Pavement roughness is not only related to safety and comfort but also a vital index to evaluate highway performance. Therefore, it is of great significance to develop a prediction model for pavement roughness. The International Roughness Index (IRI) is the most commonly used index to represent pavement roughness in the worldwide. Many deterministic models based on time series data have been developed, but there are still few researches based on panel data to study IRI. This paper aims to investigate the rich information of panel data and develop a prediction model of IRI for semi-rigid asphalt pavements in subtropical monsoon climate. With the data from the PMS of Guizhou Province, the data matrix was built from 2017 to 2020. For the independent variables, the internal and external factors affecting IRI were considered, including climate, road age, traffic, and pavement structure. The random-effects model was employed for panel data analysis and the obtained model yielded a coefficient of determination (R2) value of 0.8858. This model can be applied to the highways which are designed, constructed, and qualified under the same standard and are located in the subtropical monsoon climate region. The predicted values in 2021 were compared with the measured values of highways in Guizhou Province. The average error is 12.7%. It shows that the model has better prediction accuracy and generalization performance with good robustness to the interference of geographic region and climate.