Integrated Estimation of Stress and Damage in Concrete Structure Using 2D Convolutional Neural Network Model Learned...

Area of Science:

• Civil Engineering
• Materials Science
• Structural Health Monitoring

Background:

• Ensuring the safety and performance of concrete structures necessitates accurate stress and damage estimation.
• Capsule-like smart aggregate (CSA) technology shows promise for early detection of internal concrete damage.

Purpose of the Study:

• To propose a 2D convolutional neural network (CNN) model for integral stress and damage estimation in concrete structures using CSA sensors.
• To develop a method that learns electromagnetic impedance (EMI) responses from CSAs to assess structural conditions.

Main Methods:

• A theoretical framework for the CSA-based EMI damage technique under compressive loading was established.
• A 2D CNN model was designed to extract damage-sensitive features from CSA's EMI responses.
• Compression experiments on CSA-embedded concrete cylinders were conducted to record stress-damage EMI responses.

Main Results:

• The 2D CNN model successfully learned damage-sensitive features from CSA's EMI responses.
• The model demonstrated the capability to estimate stress and identify damage levels in concrete structures.
• The model's feasibility was validated under noisy conditions and with untrained data.

Conclusions:

• The developed 2D CNN model can simultaneously and accurately estimate both stress and damage status in concrete structures.
• This CSA-based EMI technique offers a reliable approach for structural health monitoring.
• The findings support the use of smart aggregates and AI for advanced concrete structure assessment.