提出了一种基于支持向量机超声波在线检测缺陷识别方法。首先采用小波包分析来提取超声信号的特征信息，产生训练和测试样本；然后利用支持向量机分类方法对缺陷进行识别。实验结果表明，支持向量机能够快速、有效地识别缺陷，比人工神经网络具有更好的分类性能和推广能力，是一种有效的超声缺陷识别方法。关键词：超声波检测；小波包分析；支持向量机；缺陷识别Abstract：A novel automatic identification system of flaws is presented in this paper. Wavelet packet Analysis is applied to feature extraction of ultrasonic signals, and the support vector machine is employed to perform the identification task. The experimental results suggest that the support vector machine processes better classification performance and spreading potential than neural network under the small study sample condition.Key words：Ultrasonic Testing, Wavelet Packet Analysis, Support Vector Machine, Flaw Identification