Autonomous Machine Learning Algorithm for Stress Monitoring in Concrete Using Elastoacoustical Effect

Area of Science:

• Civil Engineering
• Materials Science
• Acoustics

Background:

• Accurate stress measurement in concrete structures is crucial for safety and maintenance.
• Existing methods for concrete stress assessment can be complex and invasive.

Purpose of the Study:

• To introduce a new non-destructive measurement system, the self-acoustic system (SAS), for concrete stress.
• To demonstrate the elastoacoustic effect for correlating acoustic wave propagation with stress.
• To develop an autonomous system for real-time stress condition identification in concrete.

Main Methods:

• Utilizing a self-acoustic system (SAS) with ultrasonic heads and a positive feedback loop to achieve a stable limit cycle.
• Measuring the frequency of the limit cycle, which is related to acoustic wave propagation time and thus stress.
• Applying a machine learning algorithm for real-time analysis of the SAS system's frequency spectrum.

Main Results:

• The SAS system effectively couples limit cycle frequency with the stress degree in concrete structures.
• A real-time classifier was developed for online analysis of the SAS frequency spectrum.
• An autonomous system for stress condition identification in concrete was successfully built and described.

Conclusions:

• The proposed self-acoustic system offers a novel approach for non-destructive concrete stress monitoring.
• The integration of machine learning enables autonomous and real-time stress assessment.
• This technology has the potential to enhance the structural health monitoring of concrete infrastructure.