Vibration Signal Augmentation Method for Fault Diagnosis of Low-Voltage Circuit Breaker Based on W-CGAN

Abstract

Low-voltage circuit breaker (LVCB) fault diagnosis based on artificial intelligence (AI) algorithm has always been a research hotspot and got some recent advances. However, AI algorithms usually require sufficient data to train the model, so intelligent fault diagnosis is a challenging task when lack of fault signals. To solve this problem, a fault vibration signal augmentation method based on Wasserstein distance (WD) and conditional generative adversarial networks (CGANs) is proposed in this article. The proposed method uses WD to optimize the adversarial training of generator and discriminator, and thus, the generator can generate vibration signals under different fault conditions, which can be used to extend the training dataset. In order to verify the improvement effect of this method on the accuracy of LVCB fault diagnosis, multiple fault classifiers are trained using generated and real fault signals, and a multidimensional evaluation index system is built to evaluate the classification effect. Experimental results reveal that the method can generate fault signals with high similarity and improve the accuracy of fault diagnosis.