PSI - Issue 84

Available online at www.sciencedirect.com

ScienceDirect

Procedia Structural Integrity 84 (2026) 240–247

© 2026 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the Conference Keywords: Structural Health Monitoring; Model Updating; Reduced-Order Models; Deep Reinforcement Learning; Existing Bridges. Abstract The continuous monitoring of existing bridges is a critical task for transportation agencies, which must assess the structural health of assets within a road network to prevent potential failures. To maximize the benefits of monitoring campaigns, it is often necessary to establish a calibrated numerical model of the structure, which should be capable of capturing structural changes due to damage or aging over time. This requires a model updating process, wherein uncertain parameters, such as material properties, boundary conditions, or internal constraints, are adjusted to align numerical predictions with experimental data. However, traditional model updating procedures, which rely on Full Order FE models, can become computationally expensive and highly reliant on expert input, particularly when the number of uncertain parameters is large. To address these challenges, this paper introduces a novel model updating framework that combines reduced-order modelling with deep reinforcement learning (DRL). The core idea is to train an intelligent agent capable of autonomously tuning a high-dimensional parameter space. A reduced-order representation of the structural model is first developed to significantly lower the computational cost of training the DRL agent while preserving key dynamic features. The proposed approach is detailed, its advantages and limitations are discussed, and its performance is demonstrated on a real-life reinforced concrete bridge. III Fabre Conference: Existing Bridges, Viaducts, and Tunnels: Research, Innovation, and Applications Model updating for bridge structures using reduced-order models and deep reinforcement learning Gianluca Bruno a , Fabio Parisi a , Sergio Ruggieri a *, Eleni Chatzi b , Giuseppina Uva a a DICATECH Department, Polytechnic University of Bari, Bari, Italy b Department of Civil, Environmental and Geomatic Engineering, ETH Zurich, Stefano-Franscini-Platz-5, Zurich, 8093, Switzerland

* Corresponding author. E-mail address: sergio.ruggieri@poliba.it

2452-3216 © 2026 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the Conference 10.1016/j.prostr.2026.06.032