基于RBF神经网络与LMD的滚动轴承故障诊断方法

Fault Diagnosis of Rolling Bearing Based on RBF Neural Network and LMD

直升机传动系统故障诊断及预测对提高其运行时的可靠性和安全性具有重要意义。本研究首先采用小波包降噪与局部均值分解相结合的方法提取滚动轴承故障特征,其次用故障样本对设计好的RBF(Radial Basis Function Neural Net-work,简称RBF)诊断网络进行训练,最后利用训练好的RBF网络实现故障的智能诊断。实验结果验证了该方法能够有效地对滚动轴承故障进行分类识别。

The fault diagnosis and prediction of helicopter driving system can improve its reliability and safety significantly. In this research, the fault diagnosis method composed of wavelet packet, local mean decomposition （LMD） and radial basis function neural network （RBF） is adopted. Firstly, the noise is wiped off from the signal by wavelet packet and the treated signal is decomposed by LMD; Then, RBF is trained by using the extracted feature vector; Finally, the faults of rolling are diagnosed by the treated RBF neural network. The reasearch results indicate that the presented method can effectively identify the rolling fault.