Purdue University Graduate School
Browse

File(s) under embargo

Reason: Patents and publications under review.

1

year(s)

6

month(s)

7

day(s)

until file(s) become available

Novel Acoustic Sensing Method for In-situ Concrete Mechanical Properties Monitoring

thesis
posted on 2023-11-30, 22:09 authored by Zhihao KongZhihao Kong

In this research, a novel acoustic sensor with a waveguide is made to induce the local volumetric resonance of concrete material. The sensor is embedded in fresh concrete and monitors the in-place elastic modulus and strength development of the concrete. The resonant peak of the EMI spectrum of the sensor is governed by the concrete material in the proximate area of the sensor. The sensor itself does not affect the position of the resonant peak.

This research covers theoretical demonstration, sensor design and prototyping, remote testing systems, experimental study, and machine learning. Current work demonstrated the sensor successfully produced the resonant peaks that are related to the concrete curing process (R-square=0.86 for lab testing and R-square=0.64 for field testing); however, the sensitivity (S=1.00 Hz/psi) of the resonant frequency is not sufficient for practical application.

Machine learning algorithms were employed to map the EMI spectra to concrete strength profile. Several existing architectures were explored and evaluated. A novel machine learning scheme was proposed and successfully improved the accuracy of prediction. The algorithm is also able to handle real-time data with decent generalization among diverse concrete mixtures.

The integration test for the sensing system, including the sensor, the data collection device, the data pipeline, and the trained machine learning models, was performed in field testing of eight States. The averaged MAPE of the field prediction results is 23.43% for field structures and 16.13% for companion beam samples.

The knowledge produced during this study further advanced the application of EMI sensors in the NDE of concrete material. The EMI resonator tailored for local structural resonance is reported in this study for the first time. The EMI data processing algorithm using machine learning that is generalizable among various concrete mixtures is employed in this study for the first time. This study would be helpful for the real-world application of the EMI technique in the NDE of concrete and other phase-changing materials.

Funding

JTRP 4513

History

Degree Type

  • Doctor of Philosophy

Department

  • Civil Engineering

Campus location

  • West Lafayette

Advisor/Supervisor/Committee Chair

Na (Luna) Lu

Additional Committee Member 2

Guang Lin

Additional Committee Member 3

Yining Feng

Additional Committee Member 4

Yiheng Feng

Additional Committee Member 5

Akanshu Sharma