基于双树复小波分解的风机齿轮箱故障诊断
Fault diagnosis of wind turbine gearbox using dual-tree complex wavelet decomposition

摘要 风机齿轮箱振动信号具有非平稳、非高斯特性，多种模式混叠和复杂的传递路径使得故障信息微弱完全淹没在噪声之中.针对故障特征提取的难题，将双树复小波变换引入振动信号分析，提出了一种新的工业风力发电机齿轮箱故障诊断方法.首先对风机齿轮箱振动信号进行双树复小波分解，然后计算各频带分量的峭度值，利用峭度筛选故障敏感分量.最后对故障敏感分量进行频谱分析提取故障特征频率.实验结果表明：双树复小波变换可将复杂信号分解为不同频带分量，抑制平移敏感性和频率混叠.与传统离散小波变换相比，能有效抑制虚假频率出现并准确提取故障特征.本文提出的方法已成功用于风力发电机工业运行监测并准确诊断多种类型的齿轮箱故障.
Abstract：The vibration signal of wind turbine gearbox is non-stationary and non-Gaussian. Multiple modes are mixed and transmission paths are complicated, which make fault information weakly conceived and completely masked by noise. To solve the obstacle of fault feature extraction, a new fault diagnosis method is presented for industrial wind turbine based on the introduction of dual-tree complex wavelet transform (DT-CWT) into vibration signal analysis. First, the vibration signal of wind turbine gearboxis is decomposed by dual-tree complex wavelet analysis and the kurtosis value of each frequency sub-band is calculated.Then the fault sensitive component is selected in view of the kurtosis value. Finally, the power spectrum analysis of the fault sensitive component is performed to extract the fault characteristic frequency. Experimental results show that the dual-tree complex wavelet can decompose complex signals into different frequency bands with the advantages of reducing time-shift sensitivity and frequency aliasing. Compared with the traditional discrete wavelet transform,DT-CWT analysis could suppress the false frequency and extract fault feature accurately.The proposed method is effectively applied to the operation monitoring of industrial wind turbine and the faults of various gearboxes are diagnosed successfully.