A systematic review of rolling bearing fault diagnoses based on deep learning and transfer learning: Taxonomy, overview, application, open challenges, weaknesses and recommendations

Rolling bearing fault detection is critical for improving production efficiency and lowering accident rates in complicated mechanical systems, as well as huge monitoring data, posing significant challenges to present fault diagnostic technology. Deep Learning is now an extraordinarily popular research topic in the field and a promising approach for detecting intelligent bearing faults. This paper aims to give a comprehensive overview of Deep Learning (DL) based on bearing fault diagnosis. The most widely used DL algorithms for detecting bearing faults include Convolutional Neural Network, Recurrent neural network, Autoencoder, and Generative Adversarial Network. It discusses a variety of transfer learning architectures and relevant theories while summarises, classifies, and explains several publications on the subject. The research area’s applications and problems are also addressed.