An Evaluation Method for Pavement Maintenance Priority Classification Based on an Unsupervised Data-Driven Multidimensional Performance Model

To ensure pavement maintenance funds are reasonably allocated, it is necessary to comprehensively and effectively rank pavement sections based on multidimensional condition data to determine priority for rehabilitation. This paper proposes an unsupervised multidimensional performance data-driven model for evaluating road maintenance priority based on comprehensive multidimensional indicators, which solves the issues of a single decision index and inflexibility of traditional models. In this hybrid model, the density-based spatial clustering of applications with noise (DBSCAN) algorithm is first used to cluster the similar pavement performance properties and identify noise in the data, then the support vector machine (SVM) is employed to assess the rank and priority of different pavement section clusters. Model’s effectiveness was verified through a case study of a highway in Guangdong Province, China, in which data from 290 pavement sections were ranked according to five attributes. Six clusters and five maintenance levels were obtained, and the model was used to evaluate overall pavement performance and assign specific rehabilitation measures to different clusters. The results confirm that the proposed hybrid model requires almost no human intervention and can contribute to big data approaches to road maintenance management.