基于小波神经网络的FRP复合材料损伤声发射信号识别

摘要

针对FRP复合材料损伤声发射信号的特点,运用小波包分解提取特征向量作为网络输入,通过小波分析与神经网络紧致结合的方式对不同类型的损伤模式进行识剐,并将遗传算法引入到小波神经网络中,优化网络初始权值,提高了网络的全局搜索与识别能力.%Considering features of the acoustic emission(AE) signals from the damaged FRP composite material, the wavelet packet decomposition was employed to take characteristic vector as a network input, and then basing on wavelet analysis and combination with neural network, the damages were distinguished; in addition,the genetic algorithm was introduced to the wavelet neural network to optimize its initial weights and improve its global search and recognition capabilities.