Induction motors play a very important role in modern industry.However,motors can become tired after running for a long time,leading to catastrophic consequences.Since the essence of motor fault diagnosis is to classify the time signal of the motor,this study proposes a dual-channel Transformer model,which uses current and vibration signals to diagnose,and extracts frequency domain features through continuous wavelet transform as inputs.The dual-channel Transformer model passes the time-domain and frequency-domain signals of the data through the Transformer model respectively.This substitution can extract not only temporal characteristics,but also spatial characteristics.Experimental results show that the proposed model can provide a diagnosis accuracy of up to 95.36%,proving its effectiveness in motor fault diagnosis.Compared with traditional single-signal fault diagnosis methods,this model has better robustness and accuracy.
motor fault diagnosisdual-channel Transformer modelwavelet transformmulti-dimensional signalfrequency domain characteristics
万方数据
维普
