飞机疲劳开裂声发射波形信号的人工神经网络模式识别方法

Pattern Recognition of Aircraft Fatigue Cracking Based on Waveform Analysis Method and Artificial Neural Networks of Acoustic Emission Signals

摘要: 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(168.5 kHz)与先前的试验数据(175.8 kHz)相接近，已具备了较明显的裂纹特征。

Abstract: In this paper, SOM neural network was used to identify the AE waveform signals of aircraft fatigue test. A group of suspected crack signals were acquired. Their characteristics were obtained. Three peaks appear simultaneously in frequency spectrum. Their energies were relatively large and located at same frequency. The frequency of third peak(168.5 kHz) was consistent with previous result(175.8 kHz), and already showed obvious characteristics of crack signal.