Abstract:Aiming at the complex vibration signals in a gearbox and the difficulty in extracting fault features at the early stage of fault, a gearbox fault feature extraction and fault pattern recognition method based on adaptive Variational Mode Decomposition (VMD) and Convolutional Neural Network (CNN) was proposed in this paper. Firstly, the vibration signals of the gearbox were decomposed using VMD optimized by Krill Herd Algorithm (KHA). Then, the effective modal components are selected by kurtosis criterion for reconstruction. Finally, the reconstructed signal is used as the input of CNN for fault modal identification. The experimental results show that the proposed method is more accurate than the traditional fault pattern identification method.