基于模糊神经网络和数据融合的结构裂纹故障诊断

Fault diagnosis to structural crack based on FNN and Data Fusion

本文利用模糊神经网络和数据融合技术相结合,对铝板裂纹进行了故障诊断。在梁结构的振型斜率变化率的基础上,得到了板结构两个方向的振型斜率变化率裂纹诊断指标。通过对模糊神经网络诊断结果和模糊神经网络数据融合诊断结果的比较,证明模糊神经网络数据融合诊断方法对裂纹的检测、定位和性质刻画能够达到满意的精度,也证明了在所取五种裂纹诊断指标中,振型斜率变化率裂纹诊断指标对裂纹的诊断精度是最高的。

The fault of aluminum crack is diagnosed based on Fuzzy Neural Networks (FNN) and Data Fusion (DF). A diagnosis feature of mode slope varying ratio of plate structure in two directions is gotten from that of the beam structure. Compared with diagnosis results of both FNN and FNN and DF, it is proved that crack diagnosis based on FNN and DF could satisfy the precision of crack detection for location and length. Meanwhile, it is proved to be correct to diagnose structural crack with the mode slope varying ratio.