In this paper, we address the problem of automatic inspection of CFRP with complex surface using an ultrasonic technique. The 3D surface data are obtained by ultrasonic measurement, and then the inspection path is planned after the CAD model has been reconstructed. Defect position and size are figured out by analyzing C-Scan image. Characters of defect type are modeling according to A-wave data. Thereafter, an algorithm based on Multi-SVM is presented to classify defect types which use the energy character of defect dynamic waveform. Finally, application experiments are conducted to verify the validity and superiority of the method proposed in this paper.