Towards Intelligent Interpretation of Low Strain Pile Integrity Testing Results Using Machine Learning Techniques

Area of Science:

• Geotechnical Engineering
• Non-Destructive Testing (NDT)
• Machine Learning Applications

Background:

• Low Strain Pile Integrity Testing (LSPIT) is a widely used, cost-effective NDE method in pile foundation construction.
• Current LSPIT signal interpretation relies on manual analysis by experienced experts, leading to delays in reporting for large projects.
• Automated interpretation techniques are needed to accelerate LSPIT turnaround times and enhance efficiency.

Purpose of the Study:

• To develop a Computer-Aided Reflectogram Interpretation (CARI) methodology for automated LSPIT signal analysis.
• To assist geotechnical experts in both qualitative and quantitative interpretation of LSPIT data.
• To improve the efficiency and consistency of pile integrity assessment in deep foundation construction.

Main Methods:

• Development of a CARI methodology integrating advanced signal processing and machine learning.
• Implementation of CARI for rapid screening of numerous LSPIT signals.
• Utilizing CARI to identify potentially defective piles for focused expert review.

Main Results:

• The CARI methodology demonstrated effectiveness in interpreting LSPIT signals from real-world construction sites.
• The system can quickly screen a large volume of test signals, significantly reducing manual interpretation burden.
• CARI successfully identifies suspect piles, allowing experts to concentrate on critical cases.

Conclusions:

• The proposed CARI methodology offers a valuable tool for automating LSPIT signal interpretation.
• This approach can substantially enhance the efficiency and effectiveness of quality control in deep foundation construction.
• CARI has the potential to make LSPIT an even more powerful NDE method for infrastructure projects.