Skip to main content

2024 | Book

Long-Term Structural Health Monitoring by Remote Sensing and Advanced Machine Learning

A Practical Strategy via Structural Displacements from Synthetic Aperture Radar Images


About this book

This book offers an in-depth investigation into the complexities of long-term structural health monitoring (SHM) in civil structures, specifically focusing on the challenges posed by small data and environmental and operational changes (EOCs). Traditional contact-based sensor networks in SHM produce large amounts of data, complicating big data management. In contrast, synthetic aperture radar (SAR)-aided SHM often faces challenges with small datasets and limited displacement data. Additionally, EOCs can mimic the structural damage, resulting in false errors that can critically affect economic and safety issues. Addressing these challenges, this book introduces seven advanced unsupervised learning methods for SHM, combining AI, data sampling, and statistical analysis. These include techniques for managing datasets and addressing EOCs. Methods range from nearest neighbor searching and Hamiltonian Monte Carlo sampling to innovative offline and online learning frameworks, focusing on data augmentation and normalization. Key approaches involve deep autoencoders for data processing and novel algorithms for damage detection. Validated using simulated data from the I-40 Bridge, USA, and real-world data from the Tadcaster Bridge, UK, these methods show promise in addressing SAR-aided SHM challenges, offering practical tools for real-world applications. The book, thereby, presents a comprehensive suite of innovative strategies to advance the field of SHM.

Table of Contents

Chapter 1. Pioneering Remote Sensing in Structural Health Monitoring
Recently, long-term structural health monitoring (SHM) of civil structures by using the technology of remote sensing has received increasing attention by civil engineers. This is because such technology can facilitate SHM by introducing useful products such as synthetic aperture radar (SAR) images acquired from some satellites suitable for monitoring of large-scale civil structures in wide areas. In contrast to conventional contact-based and next-generation vision-based sensors, a long-term monitoring program via space borne remote sensing cannot provide high-dimensional structural responses. On this basis, it is feasible to conduct the program with a few SAR images. Using such products, one can extract structural responses in terms of displacements at different areas of a civil structure and monitor the responses for detecting any abnormal change. Because the long-term monitoring process is based on analyzing structural displacement responses, the main focus is on machine learning. For this process, environmental and operational changes seriously affect the performance of data-driven machine learning-aided techniques. Due to the importance of SHM in every society, this chapter intends to explain the main parts of remote sensing-based health monitoring of civil structures through SAR images and machine learning.
Alireza Entezami, Bahareh Behkamal, Carlo De Michele
Chapter 2. Advanced ML Methods: Bridging SAR Images and Structural Health Monitoring
The classical SAR-aided SHM directly analyzes displacement data obtained from SAR images. Although the extraction of displacement responses from SAR images via various interferometric techniques is of paramount importance to remote sensing-based SHM, the direct use and analysis of SAR-based displacement responses for monitoring and damage assessment of complex and critical civil structures is not a reliable and effective approach, especially in long-term monitoring with various uncertainties and variability in measured data. On this basis, this chapter proposes various machine learning methods in terms of unsupervised learning for SHM under environmental and operational conditions and different types of structural displacement responses retrieved from SAR images in terms of their sizes. In case of a large set of displacement data, an innovative probabilistic unsupervised learning method is proposed. For a small set of displacement data, this chapter proposes two unsupervised hybrid learning methods and mainly consists of three parts of data augmentation for expanding the size of small displacement responses, data normalization for removing the effects of environmental and operational conditions, and anomaly detection.
Alireza Entezami, Bahareh Behkamal, Carlo De Michele
Chapter 3. Simulating Reality: Numerical Assessments of a Bridge Health Monitoring
In contrast to conventional sensing technologies, the preparation and extraction of displacement data from a large number of SAR images are not trivial. Notably, this process is difficult and time-consuming for long-term monitoring. For this issue, this chapter aims to validate the proposed probabilistic unsupervised learning method via simulated displacement responses of a numerical model of a real-world bridge structure. On this basis, the numerical model of this bridge was developed in MATLAB environment. To simulate remote sensing-based SHM, a few points in the bridge deck are selected as the target areas for extracting structural responses. Three damage scenarios in the vicinity of one the bridge piers are simulated to assess whether the proposed probabilistic unsupervised learning method can effectively detect such scenarios.
Alireza Entezami, Bahareh Behkamal, Carlo De Michele
Chapter 4. From Theory to Reality: Advanced SHM Methods to the Tadcaster Bridge
Experimental validation of SHM methods is a critical process in evaluating the accuracy and reliability of such methods for damage assessment, especially in long-term monitoring programs. In this chapter, real-world applications of the proposed unsupervised learning methods (i.e., HMC-DAE-MD, HMC-UTSL-MD, HMC-DTL-MD, HMC-ODTL-EMD and SLS-ODTL-EMD) developed for coping with the limitation of small data are investigated by using a small set of displacement responses of a masonry bridge called the Tadcaster Bridge located in Tadcaster, UK. Because this bridge experienced a partial collapse, it is an appropriate case study for verifying the proposed methods. This chapter also indicates how a data augmentation process can help to better observe the EOCs in augmented displacement responses.
Alireza Entezami, Bahareh Behkamal, Carlo De Michele
Chapter 5. Conclusions and Prospects for Structural Health Monitoring
In this book, it has been attempted to address some major challenging issues related to long-term health monitoring of civil structures based on the technology of remote sensing and machine learning. For this reason, various unsupervised learning methods have been proposed to detect any abnormal conditions in civil structures under unknown EOCs using small and large sets of displacement responses obtained from SAR images. This chapter aims to mention the main conclusions of the proposed methods in both numerical and experimental validation stages. Due to the importance of SHM and the benefits of remote sensing, some suggestions are given to further evaluate in future research activities.
Alireza Entezami, Bahareh Behkamal, Carlo De Michele
Long-Term Structural Health Monitoring by Remote Sensing and Advanced Machine Learning
Alireza Entezami
Bahareh Behkamal
Carlo De Michele
Copyright Year
Electronic ISBN
Print ISBN

Premium Partners