Deep Learning for Structural Health Monitoring of Pavements for Improving Road Maintenance and Management

Azodinia, Mohammadreza and Kazizat, Iskakov and Orman, Indira and Faizollahzadeh Ardabili, Sina and Nader, Karballaeezadeh and Mosavi, Amirhosein (2025) Deep Learning for Structural Health Monitoring of Pavements for Improving Road Maintenance and Management. DECISION MAKING: APPLICATIONS IN MANAGEMENT AND ENGINEERING, 8 (2). pp. 863-880. ISSN 2560-6018

Preview

Text Final_DMAME-1290-20252revised1proof11.pdf - Published Version Download (922kB) | Preview

Abstract

Durable and secure roadway networks underpin sustainable mobility systems and urban growth. In recent years, Artificial Intelligence models have increasingly been employed to estimate the structural condition of roads and pavements. Such approaches enable agencies to organise maintenance schedules more effectively and to channel resources based on evidence-driven priorities. When combined with continuous sensor streams and traffic information, these systems support anticipatory and financially efficient road management, thereby strengthening decision reliability. This investigation develops machine learning frameworks to evaluate pavement structural condition. Variables including asphalt temperature and pavement layer depth are incorporated to forecast overall pavement performance. The resulting model offers a cost-efficient, unobtrusive technique for assessing current pavement status and anticipating emerging defects, which subsequently aids maintenance planning and resource distribution. Linking this modelling approach with wider smart city and smart mobility platforms facilitates real time surveillance of pavement behaviour, enhances road safety, and reinforces transport networks that favour long-term sustainability. The study compares long short-term memory techniques with alternative machine learning methods. The outcomes reveal that the long short-term memory model demonstrates superior behaviour during both training and validation, showing high predictive precision and stronger generalisation when applied to unseen datasets. Consequently, it is identified as the most suitable predictor for the data employed in this research. The results indicate that transport authorities can utilise Artificial Intelligence-based structural health monitoring systems for continuous pavement evaluation. These agencies are also encouraged to integrate sensor outputs and traffic information within unified data environments. Furthermore, policy measures that advance predictive maintenance practices are essential, as they contribute to extended pavement longevity, reduced expenditure, and improved public safety.

Actions (login required)

Edit Item