<p>To address the problems of low diagnostic accuracy due to vibration characteristic coupling and the difficulty in acquiring fault samples in the domain of multi-axis bearing fault diagnosis, this paper proposes a fault diagnosis method based on multi-axis system vibration characteristic transmission simulation. By analyzing the fault transmission characteristics of multi-axis system bearings, the vibration signal is introduced into the bearings as a fault vibration source, accurately simulating the vibration signal characteristics under different fault positions, and revealing the transmission relationship of vibration in multi-axis system bearings with faults. Based on this, the Ensemble Empirical Modal Decomposition-Principal Components Analysis-Adaptive Network-based Fuzzy Inference System-based (EEMD-PCA-ANFIS) method is adopted to diagnose the adjacent bearing faults and the faults of bearings separated by one from the obtained data, and the decline of vibration characteristics during the vibration transmission process is discovered. Supplementary training is conducted on the data of six typical working conditions of the multi-axis transmission system, and the first three IMF components are selected as feature vectors to optimize the ANFIS training model. Eventually, high-precision diagnosis of inner ring faults in the multi-axis transmission system bearings and discrimination of different bearing fault locations are achieved, and the diagnostic accuracy rate is increased to 96.6%. In contrast to the existing systems, this diagnostic approach takes into full account the interference influence of the driven shaft faults in the transmission system on the motor bearings and effectively exploits the characteristic rules of bearing fault vibrations. It exhibits higher accuracy in the diagnosis of multi-axis bearing faults and offers significant theoretical and technical support for the safe operation and fault diagnosis of multi-axis transmission systems.</p>