DTCWT多尺度联合熵和CNN的行星齿轮故障诊断方法

Fault Diagnosis Method of Planetary Gear Based on Multi-Scale Joint Entropy Feature of DTCWT and CNN

行星齿轮常被应用于大型机电装备传动系统中,但是极端恶劣工况导致其故障频发,研究行星齿轮故障诊断有助于预知性维护,提高机电装备运行效率和可靠性。提出了一种结合双树复小波变换(Dual-Tree Complex Wavelet Transform,DTCWT)多尺度联合熵特征和卷积神经网络(Convolutional Neural Network,CNN)的故障诊断方法。利用相比于普通小波变换更高级的DTCWT,将行星齿轮故障激励的特征信息分解到不同的信号分量中,结合多尺度粗粒化、频谱熵和能量熵,实现多尺度联合熵特征量化,最终结合CNN实现行星齿轮故障类型识别。通过实验验证分析,证明所提出的方法识别率达到94.5%,具有较好的诊断效果。

Planetary gear is often used in the transmission system of large-scale mechanical equipment because of its many advantages,but the extremely bad working conditions results in the frequent occurrence of faults. The research of fault diagnosis of planetary gear is helpful to predictive maintenance,and that can improve the operation efficiency.A fault diagnosis of planetary gear based on multi-scale joint entropy feature of DTCWT and CNN is proposed. DTCWT,which is more advanced than general wavelet transform,is used to decompose the vibration signal of planetary gear into different signal components. Then,the combination of multi-scale analysis,spectral entropy and energy entropy is used to realize the quantification of fault feature,and finally,the fault type recognition of planetary gear is realized by CNN. The experimental results show that the recognition rate of the proposed method has 94.5%,and which has high recognition rate.