研究目的
To compare different ray propagation models for depth estimation in Ground Penetrating Radar (GPR) data and develop a novel method for efficient depth estimation without prior knowledge of medium properties or signal emission time.
研究成果
The Bi-Static model was found optimal for the 350 MHz GPR system, with the Polynomial Approximation providing an efficient depth estimation method. The novel normalization technique eliminated the need for prior knowledge of dielectric constant and time zero, achieving an average depth estimation error of 0.344% with a standard deviation of 3.11%. This approach enhances robustness and accuracy in GPR applications.
研究不足
The model assumes noise-free data, far-field propagation, constant dielectric and magnetic properties, point antennas, and precise time measurements, which may not hold in real-world scenarios. The targets were small relative to wavelength, and further work is needed for larger targets or different mediums.
1：Experimental Design and Method Selection:
The study compares theoretical ray tracing models (Bi-Static, Mono-Static, Bi-Static with Radius Information, and Polynomial Approximation to Bi-Static) with real GPR data. A polynomial regression process is used for data fitting and normalization to eliminate dependencies on unknown parameters like time zero and dielectric constant.
2：Sample Selection and Data Sources:
Data were collected from a test pit using a GSSI UtilityScan system, with 26 copper pipe targets of diameter 2.54 cm buried at depths from 0.3 to 1.1 meters in soil.
3：54 cm buried at depths from 3 to 1 meters in soil.
List of Experimental Equipment and Materials:
3. List of Experimental Equipment and Materials: GSSI StructureScan UtilityScan system (350 MHz center frequency, Bi-Static offset of 0.16 m), copper pipes, and a test pit setup.
4：16 m), copper pipes, and a test pit setup.
Experimental Procedures and Operational Workflow:
4. Experimental Procedures and Operational Workflow: B-scans were collected over the targets. For each target, a second-order polynomial was fitted to the time-of-flight curve to estimate depth and normalize data. The l2 norm error between measured and theoretical curves was calculated for model comparison.
5：Data Analysis Methods:
Polynomial regression for curve fitting, normalization to remove time zero and dielectric dependencies, and calculation of mean square error and depth estimation error using equations derived from the models.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
