Single and simultaneous fault diagnosis of gearbox via a semi-supervised and high-accuracy adversarial learning framework

Abstract

Gearboxes are the most widely used elements for transferring speed and power in many industrial machines. High-accuracy gearbox fault diagnosis is quite significant for keeping the machine working reliably and safely. Owing to various unseen faults, it is pretty challenging to realize high-accuracy intelligent fault diagnosis of gearboxes using existing methods. In addition, existing intelligent fault diagnosis methods heavily rely on a huge number of labeled samples, and the features extraction and selection are mainly done manually. In this paper, a semi-supervised and high-accuracy adversarial learning framework for the single and simultaneous fault diagnosis of the gearbox based on Generative Adversarial Nets and time-frequency imaging is proposed. The proposed method involves two parts. In the first part, continuous wavelet transform is adopted to transform one-dimensional raw vibration signals into two-dimensional time-frequency images. In the second part, the labeled and unlabeled time-frequency images are inputted into the built adversarial learning model to realize single and simultaneous fault diagnosis of the gearbox. Finally, two case studies are implemented to verify the proposed method. The results indicate that it is higher in accuracy and fewer in training steps of achieving the highest accuracy rate than other existing intelligent fault diagnosis methods in literatures. Moreover, its performance in stability is pretty good as well.