Scholarlink

<p>Integrating artificial intelligence (AI) with Structural health monitoring (SHM) represents a critical paradigm shift in ensuring the safety and longevity of aging civil infrastructure. This review synthesizes recent advancements, highlighting the transition from traditional inspection to intelligent, data-driven systems. It examines a spectrum of AI methodologies, from established machine learning algorithms like support vector machines and ensemble methods to deep learning architectures such as convolutional and recurrent neural networks (CNNs, RNNs), which enable automated feature extraction from diverse data modalities, including vibration, acoustic, and visual sensors. Key applications reviewed include automated real-time damage detection, sophisticated condition assessment, and predictive analytics for maintenance. The paper critically analyzes persistent challenges that hinder widespread deployment, including data scarcity, the confounding effects of environmental and operational variability (EOV), and the “black box” nature of complex models, noting the emergence of explainable AI (XAI) as a crucial countermeasure. Looking forward, the review charts future trajectories toward more autonomous systems through the development of Digital Twins, the deployment of intelligence at the network edge via TinyML, and the creation of robust Physics-Informed AI models. This work underscores the transformative potential of AI to create more resilient, safe, and sustainable infrastructure.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

The convergence of artificial intelligence and structural health monitoring: a comprehensive review of methodologies, advancements, and future trajectories

  • 期刊论文

  • 发表日期：

    2025年9月29日

查看全文
  • I. V. Sarma,
  • Sarit Chanda,
  • M. Srinivasa Reddy

摘要

Integrating artificial intelligence (AI) with Structural health monitoring (SHM) represents a critical paradigm shift in ensuring the safety and longevity of aging civil infrastructure. This review synthesizes recent advancements, highlighting the transition from traditional inspection to intelligent, data-driven systems. It examines a spectrum of AI methodologies, from established machine learning algorithms like support vector machines and ensemble methods to deep learning architectures such as convolutional and recurrent neural networks (CNNs, RNNs), which enable automated feature extraction from diverse data modalities, including vibration, acoustic, and visual sensors. Key applications reviewed include automated real-time damage detection, sophisticated condition assessment, and predictive analytics for maintenance. The paper critically analyzes persistent challenges that hinder widespread deployment, including data scarcity, the confounding effects of environmental and operational variability (EOV), and the “black box” nature of complex models, noting the emergence of explainable AI (XAI) as a crucial countermeasure. Looking forward, the review charts future trajectories toward more autonomous systems through the development of Digital Twins, the deployment of intelligence at the network edge via TinyML, and the creation of robust Physics-Informed AI models. This work underscores the transformative potential of AI to create more resilient, safe, and sustainable infrastructure.