Mechanical Systems and Signal Processing
Implementation of multimodal laser ultrasonic F-SAFT algorithm for cylinders based on the cylindrical phase shift migration theory
By Feiyang Sun, Haochen Liu, Chengyi Zhang and 6 others.
Abstract
Ultrasonic methods are widely used in industry to ensure the reliable operation of cylindrical components. Traditional transducers need special designs to adapt to curved geometries. In contrast, laser ultrasonic (LU) method is well-suited for this scenario because it enables remote, noncontact evaluation of components with complex geometries. To further enhance defect detectability, the frequency domain synthetic aperture focusing technique (F-SAFT) is introduced to provide intuitive defect visualization. In this study, cylindrical phase shift migration (CPSM) method, a form of F-SAFT, is implemented and systematically evaluated for LU inspection of small internal defects in high curvature cylinders. High-frequency, broadband, and multimodal ultrasonic signals are used for CPSM imaging, providing mutual validation of the defect indication from LL and SS wave reconstructed results. Polynomial fitting and adjacent wave subtraction are used to suppress the strong thermoelastic displacement produced by co-located pump and probe beams and extract weak defect echoes for CPSM imaging. The influences of practical spatial separation between the pump and probe beams are also systematically analyzed. Beam separation reduces thermoelastic interference but violates the co-located excitation and reception assumption of CPSM, causing a predictable angular shift of approximately half the separation angle. The radial position remains nearly unchanged, although image distortion increases with separation. Finite element simulation and real LU experimental data are used to validate the reliability of the proposed imaging method. Images are reconstructed for the minimum tested defect diameters of 0.2 mm in simulation and 0.6 mm in experiment. Across the tested configurations, the maximum absolute errors in buried depth are 5.60% in simulation and 4.00% in experiment.





