Statistical damage detection based on full-field covariance of circumferential scan ultrasonic measurement

Laser ultrasonic techniques (LUTs) perform an inspection based on raster scanning pattern to obtain three-dimensional (3D) ultrasonic signals for damage detection in mechanical structures. Even though the raster scan-based in LUTs provides full-field ultrasonic data with high spatial resolution, the scan process consumes substantial time and generates redundant ultrasonic data in many applications. In this paper, statistical damage detection based on the full-field covariance of circumferential scanning is proposed to accelerate the damage detection process using LUTs. A laser ultrasonic interrogation method based on a Q-switched laser scanning system was used to interrogate 3D ultrasonic signals in a 6-mm aluminum plate with four square through-thickness at four different depths. The circumferential scans at a given radius were obtained from the 3D ultrasonic wavefield and represented in a two-dimensional (2D) matrix, angle-time (θ-t) domain. The proposed method was tested at three different circumferences where the defects were located right on, outside, and inside the area of the scan circumference. The covariance matrix, Cθ, of the vector variables in θ-direction was calculated and represented as a covariance image. The covariance image of Cθ demonstrated the ability to detect the defects at these three different circumferences. Hence, the covariance map of an ultrasound circumference can facilitate the existing LUTs to determine the damage existence instead operate in raster scanning mode.