基于Hilbert-Huang变换与人工神经网络的风机故障诊断研究  

Study on Fan Fault Diagnosis Based on Hilbert-Huang Transform and Artificial Neural Network

作　　者：王磊[1] 纪国宜[1] 

机构地区：[1]南京航空航天大学振动工程研究所,南京210016

出　　处：《发电设备》2012年第2期100-104,共5页Power Equipment

摘　　要：对风机的振动信号进行Hilbert-Huang变换并得到边际谱,以边际谱中各故障频段的能量比为元素构造风机振动信号的特征向量,利用动量法和学习速率自适应改进的BP神经网络模型对风机转子不对中、轴裂纹等故障进行诊断。结果表明该诊断方法是有效的。Hilbert-Huang transform is applied to fan vibration signal and a marginal spectrum is obtained. Eigenvector of vibration signal of wind turbines is constructed by taking energy ratio of fault bands in the marginal spectrum as elements. Faults such as wind turbine rotor misalignment and axis cracks are diagnosed by momentum method and BP neural network model improved by adaptive learning rate method. The results prove this fault diagnosis method to be effective.

关 键 词：风机 HILBERT-HUANG变换 改进的BP神经网络 故障诊断 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]