Built deep neural networks using methods such as CNNs to classify the state of civil and mechanical structures for damage classification and prevention.
Rapid SHM systems that leverage the power of deep learning would help first responders prioritise structures which are most likely to fail in the aftermath of a natural calamity. This would help save lives. In this project, we achieve ML for SHM through two approaches. We denoise the noisy signal and determine the structural health of the building from it. We also determined the structural health of the structure using ML directly from the noisy relative acceleration data. For this report, we first summarize the paper upon which we based our ML architecture. Then we discuss some of the theoretical aspects of our project such as signal denoising, followed by the objectives and the description of our project. We conclude with the results and some of the limitations of the project.