A novel approach of fault diagnosis for gearbox based on VMD optimized by SSA and improved RCMDE

Gearboxes play a vital role in the power transmission of mechanical equipment, and studying fault diagnosis methods is essential to ensure the normal operation of rotating machines. Since the vibration signal of the gearbox has unstable characteristics with strong background noise, a novel approach of fault diagnosis for wind turbine gearbox based on variational mode decomposition (VMD) optimized by sparrow search algorithm (SSA) and improved refined composite multi-scale dispersion entropy (IRCMDE) is proposed in this paper. Firstly, for reducing background noise, sample signals are decomposed by the model of SSA-VMD, and the denoised signals are recomposed according to the correlation coefficient. Then, the proposed IRCMDE under a certain scale factor is calculated to extract initial feature information of the recomposed signal. In the next step, the initial features are reduced to 3 dimensions by the algorithm of the Gaussian process latent variable model (GPLVM). Finally, a support vector machine (SVM) is used to diagnose the different states of gearbox faults. Experimental and comparative experimental results from the wind turbine drivetrain diagnostics simulator (WTDDS) show that the proposed method can quickly and accurately identify the fault of gear transmission.