Fault Diagnosis of Planetary Gearbox Using Modal Multiscale Spectral Features and 1-D CNN

The single-scale fault features of original signal from wind turbine planetary gearboxes cannot describe fault information comprehensively; instead, it generally leads to low fault diagnosis accuracy. Therefore, a fault feature extraction method is proposed for wind turbine planetary gearboxes based on the modal multiscale spectral features (MMSF) and 1-D CNN. In the feature extraction, the variational modal decomposition (VMD) is applied to gearbox vibration signal, and the intrinsic mode functions (IMFs) are obtained. These IMFs are subjected to the coarse-grained processing in the time scale. The fast Fourier transform (FFT) is performed on these IMFs, and the multiscale spectral features (MSF) of IMFs’ spectrum are extracted. Furthermore, the MMSFs are normalized and placed in the designed classifier based on 1D-CNN for training, the diagnosis modal is obtained to identify the different fault. The gearbox fault data is collected from the Wind Turbine Drivetrain Diagnostics Simulator. The experimental results show that the proposed method can improve the accuracy of gearbox fault diagnosis and has good practical value to monitor the health condition of the rotating machinery equipment.