Induction motor few-label fault diagnosis method based on fine tuning generative adversarial network

The invention provides an induction motor few-label fault diagnosis method based on a fine tuning generative adversarial network, and relates to the field of motor fault diagnosis. Establishing an improved unsupervised generative adversarial network model; training the improved unsupervised generative adversarial network model by using an unlabeled sample; performing fine tuning on a discriminator of the trained improved unsupervised generative adversarial network model to obtain a to-be-trained induction motor fault classifier; acquiring a small number of labeled samples; training a to-be-trained induction motor fault classifier based on a small number of labeled samples to obtain a trained induction motor fault classifier; acquiring a vibration signal of the induction motor to be diagnosed in an operation state; through the induction motor fault classifier after training is completed, fault diagnosis is carried out on the induction motor to be diagnosed based on the vibration signal of the induction motor t