Scholarlink

<p>Subsurface sensing using electromagnetic sensors such as ground-penetrating radar (GPR) offers scientists and engineers a unique capability to nondestructively reconstruct the composition of a subsurface target region with an unknown dielectric profile. One of the subsurface sensing applications is to detect underground/subsurface anomalies for condition assessment, such as in concrete structures and their environment. Concrete structures are widely used in the majority of modern civil, transportation, and energy infrastructures, such as buildings, bridges, tunnels, and power plants. While modern concretes are becoming more and more durable and sustainable in their design and manufacturing, deterioration and degradation of concrete structures are inevitable. Furthermore, maintenance of a growing population of existing, deteriorated reinforced concrete (RC) structures and their environment for various reasons requires reliable inspection information to understand the aging and deterioration mechanism of concrete structures for relevant maintenance measures (repair or replacement). Despite the advantages electromagnetic sensors like GPR can offer, interpretation of GPR data (1D A-scan curve, 2D B-scan image, and 3D C-scan volume) still remains a challenging task, in view of unknown material heterogeneity, environmental factors, and nonstationary electromagnetic background noises. In the past decade, artificial intelligence (AI) has emerged as a competitive approach to enable efficient, automated, and accurate analysis of GPR data. The objective of this review is to present the systematic use of AI tools (such as machine learning/ML and deep learning/DL) to aid researchers to interpret GPR data for various condition assessment applications of concrete structures and their surrounding environment. Reported use of GPR data format, GPR frequencies, GPR data type, GPR bandwidths, AI algorithms, structure type, and AI model performance are summarized and compared in each application. In this review, an emphasis has been placed on the challenges in reported AI-enabled GPR applications to identify the gaps in the state-of-the-art.</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

A Review on Integrated Subsurface Sensing in Concrete Structures and Their Environment using Ground-Penetrating Radar and Artificial Intelligence

  • 期刊论文

  • 发表日期：

    2026年7月24日

查看全文
  • Maryam Abazarsa,
  • Tzuyang Yu

摘要

Subsurface sensing using electromagnetic sensors such as ground-penetrating radar (GPR) offers scientists and engineers a unique capability to nondestructively reconstruct the composition of a subsurface target region with an unknown dielectric profile. One of the subsurface sensing applications is to detect underground/subsurface anomalies for condition assessment, such as in concrete structures and their environment. Concrete structures are widely used in the majority of modern civil, transportation, and energy infrastructures, such as buildings, bridges, tunnels, and power plants. While modern concretes are becoming more and more durable and sustainable in their design and manufacturing, deterioration and degradation of concrete structures are inevitable. Furthermore, maintenance of a growing population of existing, deteriorated reinforced concrete (RC) structures and their environment for various reasons requires reliable inspection information to understand the aging and deterioration mechanism of concrete structures for relevant maintenance measures (repair or replacement). Despite the advantages electromagnetic sensors like GPR can offer, interpretation of GPR data (1D A-scan curve, 2D B-scan image, and 3D C-scan volume) still remains a challenging task, in view of unknown material heterogeneity, environmental factors, and nonstationary electromagnetic background noises. In the past decade, artificial intelligence (AI) has emerged as a competitive approach to enable efficient, automated, and accurate analysis of GPR data. The objective of this review is to present the systematic use of AI tools (such as machine learning/ML and deep learning/DL) to aid researchers to interpret GPR data for various condition assessment applications of concrete structures and their surrounding environment. Reported use of GPR data format, GPR frequencies, GPR data type, GPR bandwidths, AI algorithms, structure type, and AI model performance are summarized and compared in each application. In this review, an emphasis has been placed on the challenges in reported AI-enabled GPR applications to identify the gaps in the state-of-the-art.