 The Cien Jiao Tong University, XJTUSY, rolling bearing dataset was used to develop a deep learning technique for bearing fault diagnosis. The RMS, Cortosis, and some of frequency energy per unit acquisition period were used as health factor indicators to divide the entire life cycle of bearings into two phases, the health phase and the fault phase. This division expanded the bearing dataset and improved the fault diagnosis efficiency. Multiscale large convolutional kernels and gate recurrent unit, GRU, networks were introduced to improve the deep convolutional neural networks with wide first layer kernels, WDCNN, network model. Training and testing processes were visualized, followed by experimental validation for four failure locations in the dataset. Experimental results showed that the proposed network model had excellent fault diagnosis and noise immunity, and could accurately diagnose bearing faults under complex working conditions. This article was authored by Linsher, Xiaowisiu, Wanzhuang Wang, and others. We are article.tv, links in the description below.