风力机齿轮箱振动信号分解方法研究

Analysis on Vibration Signal Decomposition Methods of Wind Turbine Gearbox

基于风力机齿轮箱振动信号显著的非线性及非平稳性,分别采用集合经验模态分解(EEMD)、固有时间尺度分解(ITD)和经验小波变换(EWT)分解方法对齿轮箱振动信号进行处理,求取各分解方法分量信息熵并构成特征向量,然后作为支持向量机(SVM)模型的输入进行故障识别及分类。结果表明:EWT能较好地提取振动信号中的冲击成分;ITD在3种分解方法中诊断准确率最高且最稳定,最利于风力机齿轮箱故障诊断。

Based on the significant non-linearity and non-stationarity of the vibration signal of the wind turbine gearbox, ensemble empirical mode decomposition(EEMD), intrinsic time scale decomposition(ITD) and empirical wavelet transform(EWT) decomposition methods were used to process gearbox signals respectively, and the component information entropy of each method was obtained and the feature vector was formed as input parameters of support vector machine(SVM) for fault identification and classification. Results show that EWT can better extract the impact components in the vibration signal;ITD has the highest and the most stable accuracy among the three decomposition methods, which is most conducive to the fault diagnosis of wind turbine gearbox.