A Hybrid Method Applied to Improve the Efficiency of Full-Waveform Inversion for Pavement Characterization

Area of Science:

• Geophysics
• Civil Engineering
• Nondestructive Testing

Background:

• Ground-penetrating radar (GPR) is a nondestructive testing tool for pavement layer analysis.
• Estimating thin pavement layers is challenging due to overlapping reflected pulses.
• Existing methods struggle with accuracy and efficiency for thin layer characterization.

Purpose of the Study:

• To develop an efficient and accurate method for estimating pavement layer thickness and permittivity using GPR.
• To address the limitations of conventional methods in analyzing thin pavement layers.
• To improve the full-waveform inversion process for GPR data.

Main Methods:

• A hybrid method combining multilayer perceptrons (MLPs) and a local optimization algorithm was proposed.
• MLPs provide initial estimates of thickness and permittivity in the first stage.
• These estimates initialize a full-waveform inversion algorithm in the second stage.

Main Results:

• The hybrid method significantly reduced the number of objective function calculations required for inversion (30 vs. 450).
• Applied to simulated data, the hybrid method achieved high inversion precision.
• Analysis of measured data yielded a thickness estimation error of only 1.2 mm.

Conclusions:

• The proposed hybrid method demonstrates high efficiency and accuracy in estimating the thickness and permittivity of thin pavement layers.
• This approach effectively resolves the problem of overlapping pulses in GPR analysis.
• The method offers a significant improvement over conventional techniques for GPR-based pavement assessment.