基于多层窄带局部峰值因子的变桨轴承故障特征提取

Fault Feature Extraction of Pitch Bearings Based on Multi-Layer Narrow Band Local Crest Factor

针对风力发电机变桨轴承往复运转的行程短,转速低,故障信息微弱,故障诊断困难的问题,提出了一种基于多层窄带局部峰值因子的变桨轴承故障特征提取方法。首先,使用编码器零位信号对原始信号进行分割重构,去除往复换向过程中非平稳信号和冲击的影响;然后,构建多层窄带滤波器组,设计并使用谐波局部峰值因子代替传统的谱峭度作为滤波频带的评估指标,进行滤波频带的优化选择;最后,基于选取的最优滤波频带对重构信号进行带通滤波和解调,基于平方包络谱的峰值频率及其幅值提取故障特征。设计了一种非整周往复运转轴承故障模拟试验台,用于变桨轴承故障信号的模拟和多层窄带局部峰值因子方法验证,结果表明,在基于谱峭度的滤波频带选取方法无效的情况下,多层窄带局部峰值因子方法可以有效提取变桨轴承振动信号中的故障特征。

To address the problems of short stroke,low rotational speed,weak fault information and difficult fault diagnosis during reciprocating operation of wind turbine pitch bearings,a fault feature extraction method for pitch bearings is proposed based on multi-layer narrow band local crest factor.Firstly,the zero position signal of encoder is used to split and reconstruct the original signal,removing the effects of non-stationary signals and shocks from reciprocating reversal process.Then,the multi-layer narrow band filter bank is constructed,and the fault-harmonics local crest factor is designed and used instead of traditional spectral kurtosis as evaluation indicator of filter frequency band to optimize the selection of filter frequency band.Finally,the reconstructed signal is band-pass filtered and demodulated based on selected optimal filter frequency band,and the fault feature is extracted based on crest frequency and amplitude of square envelope spectrum.A fault simulation test bench for non-periodic reciprocating operation bearing is designed to simulate the fault signals of the bearings and validate the multi-layer narrow band local crest factor method.The results show that the multi-layer narrow band local crest factor method can effectively extract the fault feature from vibration signals of the bearings when the filter frequency band selection method based on spectral kurtosis is invalid.