Abstract: Two different kinds of typical defects were obtained by FSW of Al-Cu-2219 alloy at different process parameters. Compared to X-ray, laser ultrasonic was used to detect the two defects. The signals in time domain were analyzed preliminarily, and then the feature information in frequency domain was obtained with the method of PSD-welch. The energy of different frequency band in power spectrum diagram was extracted and the maximum amplitude in time domain was taken as the characteristic. A support vector machine (SVM) model was established to identify the two defects. The results shows that laser ultrasonic can detect the defects in Al-Cu alloy effectively.