【摘要】 为提升工业滚动轴承在复杂工况下的故障识别精度，提出一种基于改进的卷积神经网络（CNN）与Transformer相结合的轴承故障诊断方法。该方法利用CNN提取振动信号的局部空间特征，并通过Transformer结构建模时间序列间的全局依赖关系，实现对关键故障特征的深层挖掘。以CWRU公开轴承数据集为实验对象，与传统CNN、LSTM和CNNGRU模型进行对比。实验结果表明，所提CNN-Transformer模型具有更高的故障识别准确率，实现了对轴承故障的有效分类，具备相当的可靠性，适用于工业环境中滚动轴承的故障检测。更多还原

【Abstract】 To improve the fault recognition accuracy of industrial rolling bearings under complex working conditions, this paper proposes a fault diagnosis method based on an improved Convolutional Neural Network(CNN) combined with a Transformer architecture. The method employs CNN to extract local spatial features from vibration signals and utilizes the Transformer structure to model global dependencies within time sequences, thereby achieving deep mining of key fault characteristics. The publicly available CWRU bearing dataset is used as the experimental benchmark, and comparative studies are conducted with traditional CNN, LSTM, and CNN-GRU models.更多还原