Confirmation of Candidature - Candidate: Hamideh Zandi Zadeh Esfahani

Structures need to be examined for cracks periodically for maintenance and safety. These cracks are signs of stress on the structures and in important locations can lead to collapse. Structural health monitoring (SHM) plays a crucial role in ensuring the safety of civil infrastructure. Traditional SHM methods often rely on manual inspections, which are time-consuming, expensive, and limited in terms of accuracy. To overcome these limitations, the integration of Convolutional Neural Networks (CNN), unmanned aerial vehicles (UAVs), and Building Information Modelling (BIM) is used for accurate SHM. This approach promises to provide more precise and efficient monitoring of structures, ensuring their long-term safety and reducing risks associated with potential structural failures.

For comprehensive inspections of structures, such as buildings and bridges, UAVs provide a unique advantage by offering the ability to capture high-resolution aerial imagery. These data can be fed into CNN algorithms which represent the current trend in crack detection. CNNs are a type of Deep Learning neural network architecture commonly used in Computer Vision applications, enabling them to classify pictures and identity cracks. Deep learning methods are currently around 90% and are not high enough for widespread use. As deep learning methods use a similar structure, improving the performance of CNN classification should yield methods for identifying potential structural defects, such as cracks. Improving a CNN model for crack detection involves several strategies that can enhance its performance and accuracy. This procedure will include pre-processing and experiment with different CNN architectures. After obtaining an improved CNN model to detect cracks in bridge and building BIM can be effectively utilized for SHM by providing a comprehensive digital representation of the physical and functional characteristics of a structure including geometric information and maintenance history. By combining UAV imagery and improved CNN outputs with BIM models, we can evaluate the identified cracks in the context of the entire structures, enabling informed decision-making for maintenance.

For more information, please email the Graduate Research School or phone 0746 311088.