Physics-guided Machine Learning for Condition Assessment of Building Structures in Operational Environments

Aug 3, 2026

A doctoral thesis develops physics-guided machine learning methods to assess structural condition under operational environments. The research frames damage identification as a learning problem constrained by physical principles, aiming to improve accuracy when measurements are limited. The work evaluates learning strategies such as transfer learning, unsupervised domain adaptation, and physics-informed operator modeling, and connects the proposed approaches to vibration-based response reconstruction and structural identification. The thesis also includes validated discussion of related publications and technical contributions.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages219

Reading Time9 h 12 min

AccessFree to download

Chapter 1. Introduction

1.1. Background

1.2. Research objectives and significance

1.3. Organisation of thesis

Chapter 2. Literature review

2.1. Concepts of structural health monitoring

2.2. Vibration-based structural damage detection

Physics-guided Machine Learning for Condition Assessment of Building Structures in Operational Environments by Xutong Zhang Thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy under the supervision of A/Prof. Xinqun Zhu Prof. Jianchun Li University of Technology Sydney Faculty of Faculty of Engineering and Information Technology March 2025 Certificate of original authorship I, Xutong Zhang declare that this thesis is submitted in fulfilment of the requirements for the award of Doctor of Philosophy, in the School of Civil and Environmental Engineering, Faculty of Engineering and Information Technology at the University of Technology Sydney. This thesis is wholly my own work unless otherwise referenced or acknowledged. In addition, I certify that all information sources and literature used are indicated in the thesis. This document has not been submitted for qualifications at any other academic institution. This research is supported by the Australian Government Research Training Program. Signature: Xutong Zhang Data: 4th March 2025 Production Note: Signature removed prior to publication. Acknowledgement For the past four years, I have had the opportunity to pursue a postgraduate program at the Centre for Built Infrastructure Resilience (CBIR) at the University of Technology Sydney (UTS), for which I am extremely grateful. UTS has provided me with an excellent, competitive and supportive learning environment that has been essential to my development. I would like to express my sincere gratitude to my principal PhD supervisor A/Prof. Xinqun Zhu and my co-supervisor Prof. Li Jianchun. Their guidance, assistance and encouragement throughout my PhD program have been invaluable to me. Their technical expertise, profound knowledge, amiable character and patience have greatly contributed to my academic and personal growth. I would like to thank Mr. Yingqi Wang for his generous help with the computational coding part of my research, as well as for his encouragement and companionship regarding my entire student career and life. I would also like to thank my good friends and my co-supervisors in my research study group, who helped me a lot in various ways and brought me a great deal of fun during my time in the study. It has been a pleasure to meet and become friends with all of you. Finally, I would like to thank my family for their constant support. I would not be able to be who I am now without you; it is your encouragement and support that have helped me to overcome all the hardships I have faced during my studies. List of publications Journals: Zhang, X., Zhu, X. and Li, J. 2023 ‘Transfer learning-based structural damage detection using frequency response functions with limited measurements’, International Journal of Structural Stability and Dynamics, [https://doi.org/10.1142/S0219455425500932](https://doi.org/10.1142/S0219455425500932) . Zhang, X., Zhu, X., Wang, Y. and Li, J. 2024 ‘Structural Damage Detection based on Transmissibility Functions with Unsupervised Domain Adaptation’, Engineering Structure. [https://doi.org/10.1016/j.engstruct.2024.119142](https://doi.org/10.1016/j.engstruct.2024.119142) Zhang, X., Wang, Y., Zhu, X., and Li, J. 2024 ‘Parallel Neural Ordinary Differential Equations based Damage Identification for Building Structure’, Structural Health Monitoring, Under review. Zhang, X., Zhu, X., Wang, Y. 2024, ‘Spatio-temporal-based physics neural operators for response reconstruction and structural identification with limited sensors’, manuscript in drafting. Conferences: Zhang, X., Zhu, X., and Li, J. 2022,‘Knowledge Transfer for Structural Damage Detection using Fine-tunning based on FRF’, International Conference on Structural Engineering Research, 22-26 Nov 2022, Sydney, Australia. Zhang, X., Zhu, X., and Li, J. 2023,‘Unsupervised Knowledge Transfer for Structural Damage Detection with Limited Data’, SHMII-12:

FAQs

What problem does the thesis address?

The thesis focuses on condition assessment and structural damage detection for building structures in operational environments.

How does the work improve learning with limited measurements?

It uses physics-guided constraints and learning strategies including transfer learning and unsupervised domain adaptation to operate effectively under limited sensor data.

What types of methods are emphasized in the research?

The document highlights vibration-based damage detection, frequency response/transmissibility-based approaches, and physics neural operator or neural ordinary differential equation modeling for reconstruction and identification.

Aug 3, 2026

Upload by Clementine