PSI - Issue 78

Gianluca Quinci et al. / Procedia Structural Integrity 78 (2026) 845–851

846

Keywords: Seismic Fragility; Machine Learning; Bridge Vulnerability; Gaussian Process Regression; Nonlinear Static Analysis

1. Introduction The evaluation of the structural condition of infrastructure assets stands today as one of the most pressing challenges in ensuring the safety and long- term resilience of Italy’s vast transportation network. The national territory hosts a dense and diverse array of bridges and viaducts, many of which were constructed between the 1950s and 1980s, an era marked by the absence or underdevelopment of seismic design codes and by design practices that differ significantly from modern standards. These structures are now affected by progressive aging, increased traffic loads, seismic hazards, and in many cases, the absence of ongoing structural monitoring, all of which highlight the urgency of implementing effective risk assessment and management systems Hao et al. 2020 and Skokandi´c et al. 2022. Traditional methods, which rely on detailed inspections and refined numerical modeling, although accurate, are not easily scalable to the entire network. Given the high number of assets distributed across the country, applying these procedures nationwide would require excessive investments in terms of time, technical staff, and financial means, rendering this strategy unsuitable for large-scale screening, El-Maissi et al. 2021. As a result, there is an increasing demand for methodologies capable of supporting a fast, initial evaluation to flag potentially critical structures and to establish priorities for further intervention. In this framework, Artificial Intelligence (AI), and more specifically, Machine Learning (ML), is gaining attention as a valuable resource to enhance infrastructure management processes. By training ML algorithms on curated datasets, it becomes possible to process and correlate a wide range of input variables, such as geometric properties, material characteristics, and seismic exposure of the surrounding area. The integration of these features allows for a first estimation of seismic vulnerability without the need to perform individual numerical simulations for each structure. Several studies in earthquake engineering have already highlighted the benefits of ML in seismic risk analysis, demonstrating that such techniques can drastically reduce computational time while maintaining a satisfactory level of accuracy, Quinci et al. 2022, Quinci et al. 2025, Quinci et al. 2023. The results produced by ML models offer a practical means for classifying and ranking infrastructures according to their seismic risk. This supports stakeholders in defining maintenance strategies, scheduling inspections, or selecting retrofitting candidates, focusing resources where they are needed most. Rather than replacing established engineering procedures, ML serves as a complementary tool, improving efficiency in terms of both time and costs, and contributing to the development of a more proactive and data-driven management approach, Xie et al. 2020. The motivations that guide this research can be summarized as follows: • Public safety: improving the capability to rapidly identify the most vulnerable bridges and reduce the likelihood of critical failures during earthquakes; • Efficient resource allocation: providing tools that enable better prioritization of funds for maintenance and strengthening interventions; • Technological advancement: fostering the integration of artificial intelligence into engineering workflows, enhancing both speed and reliability of assessments; • Scalability: proposing a streamlined, generalizable methodology applicable to extensive infrastructure networks at regional and national scales. Accordingly, this study investigates a Machine Learning-based method for the preliminary seismic assessment of bridge structures. The approach is applied to a case study located in the Sicilian Region, an area of strategic relevance due to its seismic activity and infrastructure density, in order to evaluate the feasibility and practical applicability of the methodology in a real-world context. 2. Methodological Framework for ML-Based Fragility Assessment In the field of seismic risk evaluation for bridge networks, fragility curves serve as a fundamental tool to express the likelihood that a structure will exceed a specified performance threshold given a certain level of seismic intensity. When established through advanced numerical simulations, such as nonlinear dynamic analyses or pushover