Because the vibration signal of planetary gear box has non-stationary and nonlinear characteristics, it will cause interference to its early weak fault signal under complex working conditions, and the fault information can not be correctly identified. In order to solve the above problems, this paper adopts a fault diagnosis method based on variational mode decomposition and gray Wolf optimized support vector machine. Using the central frequency approximation method, the parameter K of variational mode decomposition is solved, and the correlation analysis of the decomposed intrinsic mode components (IMF) is carried out, and the IMF components are selected for signal reconstruction. The reconstructed signals are extracted for fault features and input into Grey Wolf optimized support vector machine for fault classification. The experimental results show that the fault identification accuracy of the proposed method for planetary gearbox is 99.375%.