Abstract: Composite materials are widely used in aerospace, automobile, and other industries due to significant advantages of high strength and fatigue resistance. However, composite materials often contain internal defects such as delamination and debonding, making quality inspection crucial for curved composite parts. It is difficult for traditional manual ultrasonic detection methods to inspect these defects efficiently and accurately. To address the challenge, an automatic detection technology based on wheeled ultrasonic phased array is proposed in this paper. Initially, 3D contour point cloud data of curved composite parts are collected using a laser profilometer. Subsequently, the point cloud data undergo splicing and screening processes, followed by applying a least squares surface fitting algorithm to obtain the surface equation. The surface equation is then used to plan the detection trajectory and calculate the position and orientation of each trajectory point. the position and orientation data of each trajectory point are decomposed to derive the Interpolation angles for each axis for the gantry structure and the manipulator. Finally,enabling the implementation of automatic scanning and detection. The C-scan inspection experiment of curved composite components shows that the C-scan images can clearly reflect delamination defects.