中文摘要：

将换能器耦合电信号作为识别键合质量的信息载体，研究了耦合信号的特征提取、模式识别方法及实际应用。提出了一种换能器耦合信号细化特征提取方法，该方法以三个关键键合过程即去除氧化膜、键合点塑性变形、应力扩散为依据，将瞬态耦合信号分段进行特征提取。采用主分量分析技术对提取的特征进行选择，通过BP神经网络对键合缺陷特征进行分析，有效地识别了无引线键合缺陷，预测了键合点剪切力。

英文摘要：

Using the coupling signals of PZT as the information carrier for monitoring the bonding process,the feature extraction, pattern recognition, and practical application of the coupling signals were investigated. According to the three key bonding processes of remove the oxide film, the bonding point plastic deformation,stress diffusion,a new feature extraction method based on the sub band en- velope segmentation was proposed to characterizing the transient ultrasonic coupling signals. To elimi- nate the irrelevant information of original feature variables, the principal component analysis （PCA） method was used for the feature selection. With the proposed method, the data acquired from fault bonding states were analyzed. The results demonstrate that the proposed method can identify the wire break bonding conditions effectively and predict the bond shear test strength.