Artificial Intelligence for Structural Health Monitoring: Techniques, Applications, and Future Directions

Doroudi R., Gavgani S. A. M., Lavassani S. H. H., BEKDAŞ G., NİGDELİ S. M.

Studies in Systems, Decision and Control, Springer International Publishing Ag, ss.213-276, 2026 identifier

  • Yayın Türü: Kitapta Bölüm / Araştırma Kitabı
  • Basım Tarihi: 2026
  • Doi Numarası: 10.1007/978-3-032-07738-7_12
  • Yayınevi: Springer International Publishing Ag
  • Sayfa Sayıları: ss.213-276
  • Anahtar Kelimeler: Artificial intelligence, Convolutional neural networks, Damage detection, Data reconstruction, Deep learning, Explainable AI, Faulty sensor identification, Feature extraction, Machine learning, Predictive maintenance, Remaining useful life digital twins, Structural health monitoring
  • İstanbul Üniversitesi-Cerrahpaşa Adresli: Evet

Özet

Structural Health Monitoring (SHM) is essential for the maintenance and management of modern infrastructure, ensuring safety, longevity, and optimal performance. However, traditional SHM techniques often face limitations regarding accuracy, real-time responsiveness, scalability, and adaptability. In recent years; Artificial Intelligence (AI) has significantly enhanced SHM capabilities, revolutionizing data analysis, damage detection, and predictive maintenance. AI-based approaches—particularly machine learning and deep learning algorithms—offer robust solutions for automatic feature extraction, anomaly detection, and fault prediction, enabling real-time analysis and proactive decision-making. Architectures; such as Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) facilitate precise and automated detection of structural anomalies using data from sensors, images, and videos. Moreover; AI-driven methods effectively handle sensor faults, perform data reconstruction, and estimate remaining useful life (RUL), contributing to improving system reliability and performance. Emerging technologies; including digital twins, edge computing, and UAV-based inspections, further strengthen AI integration within SHM, paving the way for intelligent infrastructure systems. Nonetheless; challenges related to data quality, model generalization, computational limitations, and interpretability remain, underscoring key areas for future research and development.