In view of the lack of existing research on high resistance connection (HRC) faults of switched reluctance motor (SRM) and the need to improve the classification and identification accuracy of multisource signal compound faults

a diagnosis method for HRC faults and bearing faults of SRM based on Convolution Transformer New SENet (Convformer-NSE) is proposed in this paper. Firstly

the experimental platform of electric vehicle drive system based on SRM is built

and the stator winding current of motor and bearing vibration signals are collected by non-invasive measurement method as HRC faults signal and bearing faults signal respectively. Secondly

the Convformer-NSE model is used to integrate the global and local information of current and vibration signals to realize the process of automatic feature extraction and classification of the original motor signals. Finally

by comparing testing results with relevant models

the proposed method can accurately identify the HRC fault phase location and bearing fault type

and the classification accuracy can reach 100%. In addition

experimental results under different noise environments show that the proposed method has good robustness and reliability.