PSI - Issue 84

F. Foria et al. / Procedia Structural Integrity 84 (2026) 645–652

647

2. The MIREB framework The MIREB (Management and Identification of the Risk for Existing Bridges) framework is part of ETS’s broader strategy to develop integrated engineering for infrastructure risk management, following the approach already adopted with MIRET for tunnels [4] [5] and MIRETS for hydrogeological risk mapping [6]. Like these frameworks, MIREB aims to create a digital ecosystem that combines monitoring, modelling, and operational decision-making, but specifically tailored to the challenges of existing bridges. In this paper, we present the most advanced component of the framework: the real-time dynamic monitoring system based on accelerometric sensors and modal identification algorithms (SSI-COV), integrated with FEM models for continuous updating and current-state verification. This solution, detailed in the following sections, forms the technological core of MIREB’s processing layer and enables the transition toward a Digital Twin of the structure. Beyond what is illustrated here, the framework encompasses ongoing developments, including automated alarm and risk threshold management, interfaces for preventive maintenance, and integration with BIM platforms and seismic early-warning systems. In line with ETS’s previous frameworks, MIREB will also incorporate inspection data alongside monitoring results, numerical analyses, and management strategies into unified digital workflows, as already implemented in MIRET and MIRETS. The goal is to close the loop between observation and simulation, transforming raw data into actionable decisions and ensuring a scalable approach aligned with evolving standards. 3. A dynamic identification algorithm based on the SSI-COV method 3.1 The algorithm In [7] an algorithm preceding the one described here was presented. It was applied to same case study of the steel bridge described in section 4, using recordings acquired over a period of two months. The first three significant frequencies were found to be approximately 5, 8, and 11 Hz. The modal shapes associated with these frequencies were all in the vertical direction. The results were applied to a FEM model updating procedure. That algorithm presented some critical issues, including difficulty in distinguishing structural frequencies from spurious ones, the absence of an automatic procedure for analysing large amounts of data; the user-dependent selection of the actual modes, as it is performed by manually inspecting the stability diagrams. In the present paper an improvement of the previous algorithm is described, whose innovative aspects can be summarised as follows: • • automatic selection of time lag and model order parameters, without user intervention; • • creation of clusters for aggregate analysis of results; Monte Carlo analysis and Gaussian Regression Process (GPR) are used to obtain admissible values for time lag and model order. Successively an automated procedure allows the selection of the most reliable modes. The whole algorithm consists of the following steps: 1. collection of data obtained from in situ recordings; 2. preliminary analysis using a rapid identification method, such as Basic Frequency Domain method, to obtain the fundamental first frequency of the structure (or even a rough estimate of it); 3. definition of the ranges for random parameter generation of model order and time lag, and selection of a fixed number of records collected at different times during the structure's lifetime; 4. Monte Carlo random generation of pairs of model order and time lag values and for each pair identification of the model parameters using SD; 5. collection of results and analysis in aggregate form through Cumulative Distribution Function (CDF) and Probability Density Function (PDF); 6. establishment of a database of frequencies with the highest multiplicity (reference values of modal parameters); • • comparison and automatic selection of the most probable modal deformations; • • automation of the process, without user intervention in the final identification process.