In practical engineering applications

the frequency of faults of a motor is extremely low. Usually there is a lack of its fault data and a serious imbalance between normal data and fault data

which poses a challenge to the data-driven motor fault diagnosis. In order to solve this problem

the paper proposes a motor fault diagnosis method based on multi-source information fusion. Firstly

the fast spectral kurtosis feature extraction method is used to convert the motor's stator current signal and vibration acceleration signal into its spectral kurtosis feature image. Secondly

a dual-channel residual neural network model is constructed and used to integrate the fault characteristics of the vibration signal and the current signal and to complete fault classification. Finally

the fault diagnosis method based on multi-source information fusion is verified with the five fault motor datasets collected by the experimental bench. The results show that in the case of serious lack of fault data

the fault diagnosis accuracy can reach more than 95%

which is much higher than the traditional data-driven fault diagnosis method. The method is also applicable to the fault diagnosis of rotational machinery.