lace 小波对采集到的滚动轴承振动信号进行相关滤波并进行功率谱变换；然后构造卷积神经网 络（CNN）框架，并引入自注意力机制（SA）和动态选择机制（DS），进而构造 SA-DS-CNN；最后利 用 SA-DS-CNN 提取功率谱特征，依据轴承不同故障状态定位相关特征信息，实现故障特征提取和故 障识别。试验结果表明：提出模型的故障识别精度更高，准确率可达 %。提出模型在不同背景噪 声干扰下的特征学习能力和故障识别准确率优于其他组合模型，具有一定的工程参考价值。 关键词：滚动轴承；故障诊断；卷积神经网络；Laplace 小波；自注意力机制 中图分类号： 文献标识码：A Rolling bearing fault identification based on Laplace wavelet filtering and SA-DS-CNN WEI Ya-hui1, GUO Ji-yuan2, GAO Fan3 1. Dept. of Mechanical and Electronic Engineering, Zhumadian Vocational and Technical College, Zhumadian 463000, China; 2. State Key Laboratory of Mechanical Transmission, Chongqing University, Chongqing 400030, China; 3. Chongqing Wasu Robot Co., Ltd., Chongqing 400714, China Abstract: Deep learning-based rolling bearing fault identification methods were vulnerable to environmental noise, a method based on Laplace wavelet filtering with SA-DS-CNN was proposed. Firstly, in order to improve the parameter selection efficiency of Laplace wa
