Guided waves-based pile foundation characterization

The mechanical properties and effective dimensions of piles are critical factors in ensuring the safety, stability, and economic viability of civil engineering structures. This paper presents a non-invasive pile characterization method that simultaneously and automatically estimates both the length and mechanical properties of a pile. The approach involves constructing a forward model for cylindrical piles based on guided wave theory and employing a genetic algorithm for the inversion process. Utilizing the spectral element method, the guided wave model generates the dispersion relation for specified physical properties. Resonance data, including resonance frequency and resonance number, are correlated with pile length and phase velocity. The inversion method's loss function is defined by linking the forward model's dispersion relation with the experimental data. Experimental results validate the effectiveness of the proposed method, showcasing its ability to automatically characterize pile properties with high accuracy. The proposed method in this paper can achieve an accuracy of approximately 95% for each parameter through inversion.