利用SOM神经网络,对分类挑选的飞机疲劳过程采集的声发射波形信号进行模式识别分析,得到一组(300个)疑似裂纹的波形信号。其特点有:频谱图上同时出现三个明显的峰值,其能量相对较大,且频率基本固定。其中,第三峰值频率(168.5kHz)与先前的试验数据(175.8kHz)相接近,已具备了较明显的裂纹特征。 Using the SOM neural network, the acoustic emission waveform signals collected by the classified aircraft fatigue process were mode identified and analyzed, and a group of (300) undisturbed waveform signals were obtained. Its characteristics are: spectrum at the same time there are three obvious peaks, the energy is relatively large, and the frequency is basically fixed. Among them, the third peak frequency (168.5kHz) and the previous test data (175.8kHz) are close, have a more obvious crack characteristics.