Pulse eddy current (PEC) testing has the technical characteristics of fast response, wide frequency range and noncontact, which is an effective means to achieve non-destructive testing of laser welding structural defects. According to the principle of PEC detection, a finite element model of PEC detection for laser welding seam defects is established. For the PEC differential response signal obtained by simulation. the six types of feature values are proposed, including peak value (P), time of peak value(TP), peak and rising curvature ratio(PRCR), fundamental amplitude (AF), amplitude ratio of fundamental and third harmonic (AFR)and marginal sprctrum peak value(MSP) based on Hilbert-Huang transform (HHT). Next, the features are calculated and analyzed, which are selected as the input of the support vector machine (SVM), and the laser welding seam defect structure defect are identified and evaluated, which provides a new way for the non-destructive testing of weld defects in the laser welding seam process.