With strong learning ability, feature classification ability, and generalization capabilities, deep learning could be used in the feature extraction and model diagnosis based on massive, multi-source and high-dimensional measurement data, with no reliance on human intervention, and it can be widely used in the field of equipment fault diagnosis. The typical models of deep learning (Deep Belief Network (DBN), Convolutional Neural Network (CNN) and Automatic Encoders (AE)) were described, and the application progress of deep learning in bearing fault diagnosis was emphasized. Finally, the problems and development trends of deep learning in the field of bearing fault diagnosis were discussed.