Bridge Monitoring Through a Hybrid Approach Leveraging a Modal Updating Technique and an Artificial Intelligence (AI) Method
-
2022-07-01
-
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final report (08/21/2021 – 08/15/2022)
-
Corporate Publisher:
-
Abstract:An early damage identification process in bridge structures may offer an opportunity to slowdown progressive failure and thus prevent catastrophic collapses. With a structural health monitoring system which allows real-time measurement of structural responses, early damage in bridge structures can be identified with proper techniques. Recently, data-driven based damage detection has become one of the principal practices. To accommodate the requirement, the project integrates two methods (i.e., a model updating technique and an artificial intelligence (AI) prediction) that can compensate for each other’s the weakness that otherwise imposed difficulty in precise real-time application of health monitoring systems. Therefore, this project leverages a mode-updating technique with high-fidelity experimental data to obtain an accurate digital model that represents an actual bridge model. The drawback of the model updating technique (i.e., high computational time) is overcome by applying an artificial intelligence algorithm such as neural networks that are known to be computationally efficient while perusing high accuracy. In this project, a pre-trained convolutional neural network is employed to conduct machine learning for damage prediction. The performances of the proposed method are assessed with various damage scenarios. The prediction accuracy of the network is 97%.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type: