A Systematic Review of Machine Learning Methods for Damage Detection in Civil Structures: From Classical Algorithms to Deep Learning Models

A Systematic Review of Machine Learning Methods for Damage Detection in Civil Structures: From Classical Algorithms to Deep Learning Models

Shaahin G. Shirazi1 Hesam Varaee2 Zahra Faryadi3

1) Department of Electrical Engineering, Faculty of Engineering, Al-Taha University.
2) Department of Civil Engineering, Faculty of Engineering, Al-Taha University,
3) Undergraduate Student in Civil Engineering, Faculty of Engineering, Al-Taha University.

IOI Code : AAHE-ABHGHF
Publication : National Conference on Civil Engineering, Design and Construction Technologies, Energy, Materials and Advanced Machinery(icaeu.ir)
Abstract :
    Damage detection in civil structures lies at the core of structural health monitoring and serves as an effective tool for improving safety and reducing maintenance costs. Given the limitations of traditional physics-based methods, machine learning has emerged as an important data-driven approach for structural damage identification. This review systematically examines the application of machine learning methods in damage detection and analyzes the types of monitoring data, preprocessing and feature extraction stages, and the role of supervised, unsupervised, and deep learning algorithms. In addition, the transition from classical methods to deep models, including convolutional neural networks and transfer learning, is discussed. Key challenges such as the scarcity of real-world data, environmental and operational variability, generalizability, and model interpretability are also addressed. Finally, future directions in this field, including real-time monitoring, integration with computer vision, transfer learning, and explainable artificial intelligence, are outlined.

Keywords : structural health monitoring damage detection machine learning deep learning anomaly detection