摘要
将混凝土超声探伤信号的小波变换进行特征分析,提取了各级小波分解信号的谱特征,最后将这些特征输入模糊前向神经网络进行训练和分类,实验表明,这种方法具有良好的效果.
Wavelet transform was used for analyzing features of the flaw signals in ultrasonic testing of concrete. Features based on power spectrum and wave shape of the decomposed signals were extracted. Finally, a fuzzy feedforward neural network classifier was used for the features. Experimental results show that the method is effective.
出处
《无损检测》
1999年第4期152-154,共3页
Nondestructive Testing
