Abstract:The bridge network is progressively aging, with an alarming proportion of bridges over 100 years. This situation engenders substantial risks to the overall reliability of transportation networks, requiring innovative methods for efficient management. Monitoring can provide a direct source of information about structural behavior generating alerts when changes occur. Real-time alerts enable effective infrastructure management and decision-making during damage or anomalous situations. However, monitoring can result in a large amount of data that is often difficult to convert into valuable information in real time. This paper presents an approach for real-time detection of abrupt damage occurrence in bridges using unsupervised anomaly detection algorithms and strain/acceleration measurements. The approach incorporates the separation of measurements into events having the same loading nature and the construction of three feature matrices based on statistical features, time-frequency features, and wavelet spectrum features. It includes the evaluation of five anomaly detection algorithms including Isolation Forest, One-Class Support Vector Machine, Robust Random Cut Forest, Local Outlier Factor, and Mahalanobis Distance. The approach is illustrated with a case study of a steel-bascule-railway bridge, that has experienced a brittle cracking event during monitoring. Results highlight the robustness of One-Class Support Vector Machine, Isolation Forest, and Local Outlier Factor algorithms in promptly detecting abrupt changes across different features. The separation of strain and acceleration data into loading-based events, coupled with the comparison of previous and new event features, provides robust feature matrices for effective damage detection. Enhanced detection and higher scores are particularly attributed to time-frequency domain features during damage occurrence. The presented approach can be used as a base on how to perform real-time anomaly detection within the context of bridge monitoring.

Damage Detection in Structures by Using Imbalanced Classification Algorithms

A Survey On Damage Detection Methods: Advantages And Shortages

Early damage detection under massive data via innovative hybrid methods: application to a large-scale cable-stayed bridge

Damage detection for structural health monitoring using reinforcement and imitation learning

Condition Assessment of Civil Structures for Structural Health Monitoring Using Supervised Learning Classification Methods

A Nonparametric Unsupervised Learning Approach for Structural Damage Detection

A decision‐tree‐based algorithm for identifying the extent of structural damage in braced‐frame buildings

A novel unsupervised real‐time damage detection method for structural health monitoring using machine learning

Data-Driven Structural Health Monitoring: Leveraging Amplitude-Aware Permutation Entropy of Time Series Model Residuals for Nonlinear Damage Diagnosis

Multiclass Supervised Machine Learning Algorithms Applied to Damage and Assessment Using Beam Dynamic Response

Damage detection of structures based on wavelet analysis using improved AlexNet

Proposed Machine Learning Techniques for Bridge Structural Health Monitoring: A Laboratory Study

Bayesian‐optimized unsupervised learning approach for structural damage detection

Non-parametric empirical machine learning for short-term and long-term structural health monitoring

An unsupervised machine learning approach for real-time damage detection in bridges

Famine Algorithm and Pseudo‐Kinetic Energy for Structural Damage Detection

Structural Damage Detection Using Ensemble Empirical Mode Decomposition, Hilbert Transform and Artificial Neural Networks.

Damage Assessment of Structures Using Modal Test Data

Quantification of Structural Damage with Self-Organizing Maps

Structural Damage Detection and Localization with Unknown Post-Damage Feature Distribution Using Sequential Change-Point Detection Method

Structural damage identification via physics-guided machine learning: a methodology integrating pattern recognition with finite element model updating