Auxiliary information guided industrial data augmentation for any-shot fault learning and diagnosis

The label scarcity problem widely exists in industrial processes. In particular, samples of some fault types are extremely rare, even worse, the instances of certain faults cannot be accessed, but they may appear in the actual process. These two kinds of challenge together can be termed as any-shot learning problem in industrial fault diagnosis. In this paper, taking the advantages of generative adversarial networks (GAN), a generative approach is proposed to tackle the any-shot learning problem, which generates abundant samples for those rare and inaccessible faults, and trains a stronger diagnosis model. To reach this, an attribute space is built to introduce the auxiliary information, which achieves the diagnosis of unseen faults and make the generated samples more resembled to the real data. Besides, an auxiliary loss of triplet form is introduced as a joint training loss term, further improving the quality of augmented data and diagnosis accuracy. At last, the performance of model is verified by the experiments of a hydraulic system, the results of which show that our model performs excellently for both zero-shot and few-shot problems.