Title:
Multi-sensor Data Collection and Fusion Using Deep Autoencoders in Condition Evaluation of Concrete Bridge Decks
Author(s):
Jinying Zhu
Publication:
Web Session
Volume:
Issue:
Appears on pages(s):
Keywords:
DOI:
Date:
3/28/2021
Abstract:
This presentation discusses a multi-sensor data collection and fusion procedure for nondestructive evaluation/testing (NDE/NDT) of a concrete bridge deck. Three NDE technologies, Ground Penetrating Radar (GPR), Vertical Electrical Impedance (VEI), and high definition (HD) imaging for surface crack detection were deployed on a concrete bridge deck. In the GPR and VEI data analysis, a neural network autoencoder was trained to remove noise and determine relationship between the two NDE methods. A condition map based on fused features of GPR and VEI data was generated. With the high-definition imaging, the location of surface cracks was detected and plotted on the condition map. Chloride concentration tests validated the GPR and VEI measurements. Analysis shows a high correlation between GPR and VEI measurements, which indicates that the VEI test might be used as an alternative to GPR. Comparison with the HD images suggests that the transverse cracks may play a major role in deterioration of reinforced concrete.