Analytics and modeling
In order to ensure the integrity of the structure, timely and accurate detection and identification of structural damage during and after an extreme event or over the lifetime of a structure is a very important for safety and economic resons. Structural damage detection and identification techniques can be generally classified into two main categories based on whether they use dynamic or static test data. This research project will use convolutional neural networks, which will focus on leveraging the capabilities of several models and toolkits, including AlexNet, Tensorflow and Pytorch. We will establish baseline performance metrics for an existing CNN-based program using a static image set of classified bolt, corrosion and crack damages. Investigate the application of machine learning methods to improve the performance of the CNN-based program on identifying damage. Investigate the performance of the modified CNN-program on new image data set that are collected real-time using a drone.
Yuan, Zhuolin, "Structural Damage Detection Using Machine Learning Techniques" (2021). Summer Interdisciplinary Research Symposium. 93.