School of Engineering Department of Civil and Environmental Engineering 92 Exploring the Potential and Limitations of Artificial Intelligence Based Structural Health Inspections Supervisor: ZHANG Jize / CIVL Student: FU Yongning / CIVL KWAN Sum Yi / CIVL Course: UROP 1100, Fall UROP 2100, Fall Artificial Intelligence is utilized in structural health monitoring systems to automatically identify potential structural issues, such as cracks and corrosion, helping to prevent significant damage and maintain long lifespan of buildings and infrastructure. A deep learning semantic segmentation model is used for crack detection. By trial and error, our model effectively recognizes cracks in new images with similar backgrounds to the training dataset. We are exploring the possibility of detecting cracks in various complex backgrounds, as well as employing a Foundational Model to enhance the detection process. Exploring the Potential and Limitations of Artificial Intelligence Based Structural Health Inspections Supervisor: ZHANG Jize / CIVL Student: KWAN Sum Yi / CIVL Course: UROP 3200, Spring Artificial Intelligence is utilized in structural health monitoring systems to automatically identify potential structural issues, such as cracks and corrosion, helping to prevent significant damage and maintain long lifespan of buildings and infrastructure. A deep learning semantic segmentation model is used for crack detection. By trial and error, our model effectively recognizes cracks in new images that resemble the training dataset. We are exploring the possibility of detecting cracks in various complex backgrounds by applying advanced techniques like Fourier Transfer and Style Injection. Exploring the Potential and Limitations of Artificial Intelligence Based Structural Health Inspections Supervisor: ZHANG Jize / CIVL Student: LO Tin Yiu / CIVL Course: UROP 1100, Spring The feasibility of utilizing artificial intelligence for the inspection of structural health was explored in this project. Given the challenges of traditional inspection methods, especially with the complexity in the urban environments with skyscrapers and high-rise apartments, this research aims to simplify structural health monitoring through artificial intelligence. Multiple approaches were evaluated, including image augmentation and Fourier transform techniques. The study found that applying the Fourier transform significantly improved model performance, achieving over 98% accuracy in detecting structural deficiencies. These findings suggest that incorporating advanced AI methods could enhance regular structural inspections, ultimately contributing to safer building practices in modern urban settings. Distributed Fiber Optic Strain Sensing for Civil Infrastructure Monitoring Supervisor: ZHANG Shenghan / CIVL Student: CHEUNG Hoi Yan / CIVL Course: UROP 2100, Fall UROP 3200, Spring This research highlights the importance of reinforced concrete structural integrity and the role of structural health monitoring (SHM) in identifying failures. In order to improve measurement accuracy and data reliability and ensure infrastructure safety, fiber optic sensors for continuous measurement were studied, with a focus on strain transfer effects.
RkJQdWJsaXNoZXIy NDk5Njg=