In order to accurately identify fault types from pipeline vibration signals, a multi-scale feature combination optimization method for aviation hydraulic pipeline fault diagnosis was proposed. The variational modal decomposition parameters were determined by using the energy ratio method to achieve the optimal decomposition of pipeline vibration signals, and the optimal modal component signals were selected for reconstruction as analysis signals. The optimized multi-scale dispersion entropy of the reconstructed signal was chosen as the characteristic indexes, a representative set of eigenvectors was constructed and input to the limit learning machine network optimized by using the sparrow search algorithm for training to realize the fault diagnosis of the aviation hydraulic pipelines. The results show that by using this method, the fault types of aviation hydraulic pipelines can be effectively and accurately identified, which provides a feasible diagnosis idea for distinguishing the fault types of aviation hydraulic pipelines.