 This study proposes a convolutional neural network, CNN, for identifying and classifying bearing faults. It replaces the commonly used max pooling layers with average pooling layers and uses a global average pooling layer instead of a full connection layer. Additionally, it employs a batch normalization layer to optimize the model. The CNN is trained using multi-class signals from two datasets, XJTUSY and Potterborn University, which demonstrate its ability to accurately identify and classify different types of bearing faults. This article was authored by Xiong Zhang, Jialu Li, Wenbo Wu, and others.