Dynamic Identification Method for Highway Subgrade Soil Compaction Based on Embedded Attitude Sensors.

Area of Science:

• Geotechnical Engineering
• Civil Engineering
• Sensor Technology

Background:

• Traditional subgrade compaction testing is destructive and provides delayed feedback.
• Ensuring long-term performance of subgrade structures relies heavily on compaction quality.
• Limitations of current methods necessitate advanced, non-destructive techniques.

Purpose of the Study:

• To develop a dynamic identification method for subgrade compaction using embedded attitude sensors.
• To establish a real-time, non-destructive approach for assessing soil compaction quality.
• To offer a technological solution for intelligent quality control in road construction.

Main Methods:

• Embedding customized sensor units with inertial measurement modules in soil samples.
• Utilizing signal processing, including improved wavelet-based denoising (TMI vs. AWT), to analyze sensor data.
• Extracting attitude features (yaw, pitch, roll angles) and applying regression models (Ridge, Random Forest, XGBoost).

Main Results:

• The yaw angle showed high sensitivity to vertical settlement during compaction.
• XGBoost model achieved R² > 0.995, indicating excellent predictive performance for compaction degree.
• Transient Masking Interpolation (TMI) effectively preserved compaction trends compared to AWT.

Conclusions:

• Embedded attitude sensors provide a viable method for dynamic subgrade compaction assessment.
• The proposed sensor-based approach offers a promising alternative to traditional destructive testing.
• This technology has strong potential for real-time field applications in intelligent road construction quality control.