Predictive Maintenance of Road Networks via Spatial Clustering and Scheduling: a Data-Driven Approach

This presentation outlines the results on prediction maintenance of road network infrastructure, as part of the research pursued within the EvoRoads project (Horizon Europe, Grant Agreement ID: 101147850). The prognostic/predictive solution comprises a data-driven framework that integrates predictive analytics, spatial clustering, and optimization techniques to support more effective maintenance strategies. The presentation, delivered at the 6th International Conference on Control and Fault-Tolerant Systems (SysTol 2025), held in Ayia Napa, Cyprus (6-8 October 2025), outlines: Tools for evaluating individual pavement condition metrics that incorporate threshold-based maintenance triggers. A density-based clustering algorithm that groups identified maintenance needs into geographically coherent clusters. A cluster-level optimisation algorithm that prioritise interventions within budget constraints. A cascaded architecture of two modules (prediction-optimisation) that enhances resource allocation, minimises risk in high-priority zones and enables more proactive and spatially-informed infrastructure management. Experimental results that highlight the effectiveness of the proposed approach. Conference official website: [click here]Speaker: Dimitrios Papageorgiou from DTU