Fault feature extraction of wind turbine gearbox based on NMD and S-ADE-VMD

As an important mechanical transmission system in wind turbines, the gearbox directly affects the working efficiency of the wind turbine. However, due to the randomness of wind speed and the harsh working environment, gearboxes often operate at high loads, making them one of the most prone components to failure in wind turbines. Meanwhile, due to the complex frequency components of gearbox vibration data, severe noise interference, and obvious abnormal characteristics, it is difficult to accurately extract failure features. Therefore, this article proposes a noise-invariant strategy based on NMD (nonlinear mode decomposition) and S-ADE-VMD (stationary signal, adaptive differential evolution, variational mode decomposition) for fault feature recognition in non-stationary turbine gearbox vibration signals. In this paper, NMD is integrated into the processing of the original vibration signal, which can adaptively decompose the components within a specific frequency bandwidth and filter out the noise interference. Then, the obtained component signals are screened to form reconstruction signals for resampling to obtain angle domain stationary signals. The angle domain signals are processed by S-ADE-VMD. Finally, the fault information is extracted. The experimental results show that the algorithm can filter out the noise components of wind turbines under different load conditions, and the features of each order are relatively obvious. It has good robustness for extracting gearbox feature information.​