基于振动的风力发电机故障诊断  

Fault Diagnosis of Wind Turbine Based on Vibration

作　　者：李刚[1] 白宇君[1] 史鸣 王新梅[1] 

机构地区：[1]兰州交通大学机电工程学院,甘肃兰州730070

出　　处：《兰州交通大学学报》2015年第4期123-126,共4页Journal of Lanzhou Jiaotong University

摘　　要：运用小波包分析采集风力发电机组工作的振动信号的去噪滤波,在多尺度分辨率的情况下提取高、低频故障特征向量,对故障特征向量进行BP网络训练,以实现风力发电机的故障诊断.首先搭建实验平台,以模拟转子不平衡、转子裂纹和油膜振荡,再利用小波包分解系数得到各频带的能量谱,最后经过BP神经网络进行检测,以达到对这些故障进行诊断的目的.The wavelet packet analysis is used to collect the denoising and filtering of wind turbine vibration signal.In the case of multi-scale resolution,the fault feature vectors of high and low fre-quency are extracted,and they are trained by BP network in order to achieve the wind turbine fault diagnosis.Firstly,the experimental platform is built to simulate the unbalance of rotor,rotor crack and oil film oscillation,then the energy spectrum of each frequency band is obtained by u-sing the wavelet packet decomposition coefficient.Finally,it is detected by BP neural network in order to discriminate the failure modes successfully.The results show that this method is effec-tive.

关 键 词：小波包分析 故障特征向量 BP神经网络 

分 类 号：TM315[电气工程—电机]