Wavelet packet decomposition (WPD) was applied to feature extraction of ultrasonic signals, chaos-genetic optimization to eliminate redundant and irrelevant features and support vector machine to perform the identification task.

• 该系统采用小波包分解提取超声信号的特征,采用混沌遗传算法来消除冗余和不相关的特征,并采用支持向量机(SVM)来对缺陷进行识别。 为了验证系统的有效性,在实验室内作了大量的超声检测实验。