| Author: | Arshad, Husnain |
| Title: | AIoT-based multi-sensing structural damage assessment for modular integrated construction (MiC) modules during transportation and assembly |
| Advisors: | Zayed, Tarek (BRE) |
| Degree: | Ph.D. |
| Year: | 2025 |
| Department: | Department of Building and Real Estate |
| Pages: | xvi, 267 pages : color illustrations |
| Language: | English |
| Abstract: | Modular integrated construction (MiC) is widely adopted by industry and governments. However, its fragile and delicate logistics are still a concern for impeding project performance. MiC logistic operations involve rigorous multimode transportation, loading-unloading, and stacking during storage. Such rigorous logistics may cause intrinsic damage to the module, leading to a safety hazard and structural deterioration during building use. Meanwhile, the inevitable supply chain uncertainties add to the complexities, challenging the just-in-time (JIT) assembly goal. Therefore, continuous monitoring of the module structure during MiC logistics and the building use phase is vital. Consequently, the objectives of this research are to (1) investigate critical factors influencing MiC logistic operations, (2) explore technologies for addressing the challenges of MiC logistic operations, (3) develop a real-time sensing system for monitoring the MiC Module, and (4) develop a deep learning model for the MiC module's damage assessment. To achieve first objective, the factors influencing the MiC supply chain and logistics are investigated through systematic review, eigenvector ranking, and MICMAC analysis. The second objective explores potential supply chain technologies and their benefits for MiC logistics using an NVIVO text analytics approach. Then, the synergies between the technologies' benefits and MiC challenges are discussed to enlighten the most beneficial technologies for MiC logistics. For the third objective, a multi-sensing IoT device is designed and developed to monitor the module's structure. The developed device is calibrated and tested to ensure accuracy. Application of the developed sensing device is also demonstrated for the MiC module's damage and safety monitoring through a detailed field experiment. A hybrid deep learning model is developed for damage assessment in the fourth objective. The model's architecture integrates the convolutional and sequential deep learning models. Model testing and validation are performed using a damage assessment scenario from the MiC field experiment. The analysis of the influencing factors revealed critical factors, their interrelationships, and the themes demonstrating the factors' influencing mechanisms. Results also highlight prevalent factors affecting the MiC supply chain and potential factors that need further research attention. The synergy analysis between technologies highlighted the most beneficial technologies and the least addressed MiC challenges. BIM, RFID, and Blockchain are widely used but still lack applications to support several other MiC challenges. One such challenge is ensuring modules' structural safety and damage monitoring during transportation and assembly. A multi-sensing IoT device has been developed to deal with the critical issue of real-time monitoring of the module structure. A compact and portable sensing device is designed to ensure its practicality for MiC modules while integrating an accelerometer, gyroscope, and strain sensors. Temperature calibration is performed using regression models to improve its accuracy. The device's performance prevails over the standardized commercial equipment, with less than a 5% difference. The application of the developed multi-sensing systems is successfully demonstrated for damage assessment on MiC modules using conventional methods, such as moving window analysis, FFT, strain histograms, etc. However, these analyses involve data pre-processing, excessive calculations, and the lack of capabilities for automated real-time assessment. The developed hybrid CNN-GRU deep learning model ensures the real-time automated damage assessment, having an accuracy (R²) of 96%, with negligible mean square error. The deep learning model prediction led to accurate damage level assessment and localization for damage case scenarios. Overall, this research theoretically contributes to the MiC, logistics supply chain, A-IoT sensors, and structural damage monitoring knowledge domains by (1) identifying the most critical influencing factors, their interrelationships, and mechanisms to influence the MiC supply chain; (2) identifying the most useful technologies for MiC logistics and highlighting the technology gap for addressing the MiC challenges; (3) integrating multiple structural response measurement sensors and wireless communication systems and establishing a robust IoT communication framework for the large data real-time transmissions; (4) evaluating the conventional damage assessment methods performance for the case of non-stationary MiC logistic operations; and (5) developing a hybrid damage prediction model architecture by integrating the convolutional and sequential deep learning models. Meanwhile, the study also offers practical contributions to the construction industry in the form of (1) a framework of critical MiC supply chain factors for improving the policies and logistic strategies, (2) identifying the most beneficial technologies available for improving MiC operations, (3) enabling the real-time module structural monitoring using the developed IoT sensing system, and (4) robust, automated damage prediction with the developed hybrid deep learning model. |
| Rights: | All rights reserved |
| Access: | open access |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/13620