High-efficiency surface defect detection based on laser ultrasonic state space embedding and compressive sensing

Abstract

Abstract The laser nonlinear wave modulation spectroscopy(LNWMS) technique has gained considerable attention due to its high sensitivity in detecting small surface defects and its ultra-fast scanning speed. This paper proposes a novel method for synthesizing intact wavefield reference, significantly enhancing the accuracy of surface defect imaging. Moreover, considering the potential for parallel processing of the nonlinearity calculation of ultrasonic signals at scanning points, we incorporate compressive sensing technology to accelerate this process. This innovative approach reduces the computational load to 10% of the original, thereby substantially increasing the imaging speed. The paper validates the method’s superior accuracy and efficiency in defect detection through conducting experiments using a high-speed laser ultrasonic scanning system on aluminum plates and turbine blade, and by comparing with local wavenumber estimation, demonstrating the promising potential of this technology for surface defect analysis.